Membrane Based Heat Exchanger

Sofie Marie Aarnes

Master of Science in Product Design and Manufacturing Submission date: June 2012 Supervisor: Hans Martin Mathisen, EPT Co-supervisor: Maria Justo Alonso, SINTEF

Norwegian University of Science and Technology Department of Energy and Process Engineering

Preface

Preface This master thesis was written at the Department of Energy and Process Engineering at the Norwegian University of Science and Technology in Trondheim, Norway through the spring semester 2012. The project was done as a part of the Zero Emission Building Research Center and in collaboration with SINTEF Energy. My supervisor was Professor Hans Martin Mathisen at EPT and co-supervisor was Engineer Maria Justo Alonso at SINTEF Energy. I want to thank both my supervisors for letting me be a part of this interesting project and for support and help through the process. I would like to thank Håvard, Lars, Magnus, Reidar and Stein at the EPTlaboratory for help in building the test rig and set up of the measurement instruments, Erik Langøren for help in writing the HSE-report as well. Carol Pionneau was working with me for two months in the beginning of the project and came up with a lot of good ideas regarding the test rig and instrumentation. I would also thank Bjarne Malvik at SINTEF for help in measuring temperature and humidity in the preface testing of the test rig. Finally I would like to thank my fellow students for good discussions, support and a lot of social coffee breaks.

Sofie Marie Aarnes Trondheim, June 2012

v

Sammendrag

Sammendrag Energibruken knyttet til ventilasjonsanlegg i boliger er et stort og viktig tema når utbredelsen av lufttette lavenergihus og passivhus øker i omfang. I boligkompleks med flere boligenheter og sentraliserte luftkondisjoneringsanlegg er det viktig å unngå lekkasjer av lukt og andre forurensinger fra avkastlufta til inntakslufta. Dette gjør at platevarmegjenvinnere er den vanlige varmegjenvinnertypen i slike anlegg. I denne oppgaven ble det sett på om en membranbasert platevarmegjenvinner som i tillegg til varme kan utveksle fukt, vil ha mindre problemer med kondensering og igjenfrysing enn en platevarmegjenvinner av plastmateriale. I tillegg ble det utviklet en matematisk metode og et enkelt regneverktøy for å simulere varme- og fuktoverføringsvirkningsgraden for en slik membranbasert varmegjenvinner. For å sammenlikne de ulike typene av varmegjenvinnermateriale ble det bygd et testoppsett i laboratoriet hos institutt for energi- og prosessteknikk ved NTNU. Eksperimentene viste at det oppsto kondens og frost i noen områder inne i de plastbaserte varmegjenvinnerne, mens i den membranbaserte gjenvinneren var det ingen tegn til verken kondenserings- eller fryseproblemer for temperaturer ned mot -10 oC i de samme områdene. Membranen hadde derimot en tendens til å utvide seg mye ved høy luftfuktighet og dette førte til at membranene klistret seg sammen. Både kondensering og tilfrysing ble observert i et at forsøkene der dette skjedde. Den testede membranen var på grunn av dette dermed ikke optimal med tanke på videre bruk i en membranbasert varmegjenvinner. Den utviklede metoden for å beregne virkningsgraden for fuktoverføring korrelerte meget godt med de eksperimentelle forsøkene. Det utviklede programmet kan dermed brukes for å forhåndspredikere andre membraners egnethet for bruk i en membranbasert varmegjenvinner.

vii

Abstract

Abstract Reduction of the energy used to acclimatise buildings is a huge challenge simultaneously with the implementation of air tight low energy buildings. In residential buildings with several living units centralised air handling units are the most energy efficient system. However, in a centralised system there is important to avoid leakages of pollutions between the exhaust air and the supply air. This leads to that flat plate heat exchangers are used instead of the more energy efficient rotary heat exchanger in these types of buildings. Flat plate heat exchangers will have problems concerning water condensation and frost formation in the exhaust air channels at low supply inlet temperatures. In this thesis a membrane based heat exchanger, which also was able to transfer moisture, was compared to a plastic based heat exchanger to see if the membrane based exchanger had less problems concerning condensation and freezing. In addition a mathematical method was derived to predict the heat and moisture transfer effectiveness in a membrane based heat exchanger. To compare the different heat exchanger plate materials a test rig was built in the laboratory at the Department of Energy and Process Engineering at NTNU. The experiments showed that the plastic based heat exchangers had problems with condensation and freezing in the tested conditions. The membrane based exchanger did not experience the same problems. However, additional problems with expansion of the membrane in high humidity showed that the tested membrane had drawbacks and was not really suitable. The derived mathematical method to predict the moisture transfer effectiveness was shown to correlate very well with the experimental results. The derived method and the developed Microsoft Excel tool called HXcalc may then be used to investigate other membranes moisture transfer effectiveness.

ix

Contents .

Contents

1

2

3

4

5

Introduction

1

1.1. 1.2. 1.3.

1 3 4

Mathematical Model

11

2.1. 2.2. 2.3. 2.4.

11 13 22 24

Membrane based heat exchanger theory Heat Exchanger Mass Transfer Practical use of the mathematical model Evaluation of the Method in HX Applications

Experimental Investigation: Method

27

3.1. 3.2. 3.3.

27 30 33

Development of the Test Rig Measurement and Instruments Uncertainty

Results

39

4.1. 4.2. 4.3. 4.4.

39 41 42 46

Results from the Mathematical Model Flow Pattern inside the Heat Exchanger Results from the Experimental Investigation Expansion of the Membrane in High Humidity Conditions

Discussion

49

5.1.

49

5.2. 5.3. 5.4.

6

Background Objective Literature

Comparison between the Mathematical Model and the Experimental Investigation Pressure Drop and Flow Rates through the Exchanger Evaluation of the Test Rig Evaluation of the Membrane Based Heat Exchanger Prototype

Conclusion and Further Work

References

51 53 54

59 61

Appendix

xi

List of Figures .

List of Figures 1.1.

Flow pattern in a quasi-counter heat exchanger. Reprint from (Zhang, 2010).

5

1.2.

The coherence between the outdoor temperature and indoor relative humidity for Finnish residential buildings. Reprint form Kalamees et al. (2009).

5

1.3.

Reprint from (Yun et al., 2002). Frost formation on a horizontal cold plate.

6

1.4.

Water molecules in the gas stream may either bond to polar groups in the polymer membrane surface (white balls) or to other water molecules already absorbed to the surface.

8

2.1.

Flow over membrane.

13

2.2.

The normal operation area for a membrane based heat exchanger in cold climate.

16

2.3.

Upper: Results from Gibson(2000): water vapour transport resistance for different types of membranes for different humidities at 3oC. Lower: Resistance equivalent values from sorption curves from (Marais et al., 2000),(Modesti et al., 2004) and (Niu & Zhang, 2000).

17

2.4.

Comparison between measured values by Gibson(2000) and empirical relations by the author.

20

2.5.

Comparison between measured values by Gibson (2000) and empirical relations by the author.

21

2.6.

User interface of the input sheet in the developed HXCalc program.

22

2.7.

Results sheet after clicking on a single point. An information window appears.

23

2.8.

Results sheet after clicking on a single point. An information window appears.

23

2.9.

Comparison between the moisture transfer effectiveness calculations methods on changing supply inlet relative humidity from Aarnes, Kadylac et al.(2009) and Niu and Zhang(2000).

24

xiii

List of Figures

.

2.10.

Comparison between the derived model and results from Niu and Zhang (2000) for changing flow rate.

25

2.11.

Comparison between the derived model and results from Niu and Zhang (2000) for changing supply inlet temperature.

25

2.12.

Required and saved energy for heating. Comparison between the derived model and results from Niu and Zhang (2000).

26

3.1.

Drawing made in Google Sketch Up.

28

3.2.

Heat exchanger made of sandwich construction. Drawing of concept in Google Sketch Up. 3 layer prototype.

29

3.3.

Die for placing of the heat exchanger. The rectangular ducts with pressure measurement pipes are shown.

30

3.4.

Measurement set up.

31

3.5.

Flow chart of the experimental set up.

32

3.6.

Random uncertainty in the pressure drop measurement. Print screen from LabView Front Panel interface.

33

3.7.

Pressure drop measurement. A linear trend line equation for exhaust air pressure drop is displayed to the right.

34

3.8.

Humidity level measurement. Chart from LabView interface.

35

4.1.

Temperature, moisture transfer and annual heat recovery effectiveness for a set point temperature at 18 oC for different air flow rates(m3/s).

40

4.2.

Simulated supply inlet temperature dependence for the efficiencies.

40

4.3.

Popup window from HXcalc for supply air temperature at -10oC.

41

4.4.

Middle streamline for three different photo series. Picture to the right shows a streamline when the paper was exposed to high humidity.

41

4.5.

Adobe Photoshop was used to overlay several pictures to see the different streamlines.

42

4.6.

Temperature efficiency for all experiments.

43

4.7.

Pressure drop for all experiments except 7 and 8.

43

4.8.

Pressure drop for experiment 7 and 8.

44

4.9.

The moisture transfer efficiency dependency on time for the different experiments with membranes.

45

xiv

List of Figures .

4.10.

The moisture transfer efficiency on the exhaust air side dependence on time for the different experiments with membranes.

45

4.11.

Ice formation in experiment (from left) 1,4 and 8.

46

4.12.

Left: Dry heat exchanger. Right: Wet heat exchanger

47

5.1.

Comparison between the experimental and simulated results for temperature efficiency (left) and moisture transfer efficiency (right).

50

5.2.

Idealistic quasi-counter flow heat exchanger flow to the left and the results from the smoke-pen test to the right.

51

5.3.

Correlation between the temperature efficiency and the mean pressure drop.

52

5.4.

Correlation between the exhaust flow rate and the exhaust side pressure drop (left). Correlation between the supply flow rate and the supply side pressure drop (right).

52

5.5.

The dimensionless temperature change and flow pattern in the warm air side through a quasi-counter flow heat exchanger. Reprinted overlaid figures from Zhang (2010).

54

5.6.

Left: The dimensionless humidity change and flow pattern in the warm air side through a quasi-counter flow heat exchanger. The exhaust air inlet is at the lower rights side. Reprinted overlaid figures from Zhang (2010). Right: the ice formation and crumpling of the tested heat and moisture exchanger.

56

5.7.

Lines drawn between the membrane based experiments inlet conditions in a temperature-humidity diagram.

57

xv

List of Tables .

List of Tables 3.1.

Properties for the different tested plate materials.

29

3.2.

Uncertainty of measurement instruments used in experimental set up.

31

3.3.

Uncertainty of parameters in manual flow measurement method. Numbers from experiment 1.

35

4.1.

Simulation input values.

39

4.2.

Overview of all experiments with mean values for inlet temperatures and relative humidity, pressure drops, measured air flow rates and calculated efficiencies based on the exhaust air side.

42

4.3.

Difference in pressure drop over time.

44

xvii

Nomenclature

Nomenclature

̇

̃ ̅

̅ ̇

Area Coefficient in the mass-heat transfer analogy Concentration Constant in resistance-humidity relation Constant in Zhang and Niu’s sorption equation Diffusivity Flow rate Heat transfer coefficient Height of air layer in permeability test. Henrys constant Hydraulic diameter Length scale Lewis number Linear trend line value Mass Mean value Moisture flow rate/flux Moisture transfer coefficient Moisture transfer Resistance Mol Molar weight Number of measurements Number of transfer units Nusselt number Overall heat transfer coefficient Overall moisture transfer coefficient Permeability Pressure Ratio between the min and max air flow Solubility Standard deviation Temperature Thickness Time Uncertainty Uncertainty Universal gas constant Volume Volumetric flow rate

xix

or mol

⁄ ⁄

Nomenclature

.

Greek letters ⁄

Absolute humidity Aspect ratio (height/width) Density Effectiveness Efficiency Moisture content in membrane Porosity Relative humidity Tortuosity Volume fraction

Subscripts Air Convection Exhaust air side Film Inlet side Maximum Membrane Minimum Moisture Outlet side Random Reference Supply air side Systematic Total Water

xx

Keywords

Keywords Absolute humidity

The amount of water vapour in air. Given as the ratio between the kilograms of water to the kilograms of air mixture.

Air handling unit

Usually the mechanical components in the air condition system for a building is connected in a single box called an “air handling unit”. Typical components are fans, filters, sound attenuators, heating and cooling elements and a heat recovery system.

Effectiveness

The same as the efficiency, but corrected for the difference in flow rates. This gives the same effectiveness for both streams and is a single number for the heat exchanger.

Efficiency

The amount of heat an air stream takes up or gives away from or to the other stream divided on the total heat difference in those two streams. The efficiency for both streams will coincide if the flow rates are equal.

Energy recovery ventilator (ERV)

Heat and moisture exchanger.

Heat recovery

Utilizing the heat in the exhaust air or waste heat from for example the discharged water in a heat exchanger to heat (or cool) air or water to the HVAC system.

Heat recovery ventilator (HRV)

Heat exchanger.

Hydrophilic

Means “water loving”. A hydrophilic material absorbs water due to its polarity.

Microporous

A material with pores usually less than 2 nm. The material lets tiny water vapour molecules pass through while bigger molecules cannot pass through.

Moisture transfer resistance

The resistance of a material to let water vapour pass through it.

NTU

Number of transfer units. A dimensionless unit used in the effectiveness-NTU method to predict the effectiveness of a heat exchanger.

Permeability

The ability of a material to let water vapour pass through it.

Relative humidity (RH)

The amount of water vapour in air. Given as the percentage ratio between partial pressure of water vapour to the saturated vapour pressure at the same conditions (temperature, pressure).

ZEB

The Zero Emission Buildings Research Centre

xxi

Chapter 1: Introduction

Chapter 1

Introduction 1.1. Background In the Nordic countries the population spend most of the time inside buildings. To feel comfortable the indoor air has to be within a specific temperature range (Novakovic et al., 2007). This is done by either heating in cold outdoor conditions or cooling in warm outdoor conditions. Acclimatisation of buildings requires a huge amount of energy per year. The energy efficiency of the acclimatisation systems in buildings have improved through the last decades. However, still more research is needed to continue the improvement. The Zero Emission Buildings Research Centre (ZEB) was initiated by the Research Council of Norway in 2009 as one of eleven national centres of Environment-friendly Energy Research (FME). Hosted by the Faculty of Architecture and Fine Art the research centre shall coordinate the collaboration between several partners ranging from manufacturing companies to architects and several different research and teaching institutions. The aim of the ZEB project is to: “Over the next eight years, the FME-Centre ZEB will develop competitive products and solutions for existing and new buildings that will lead to market penetration of zero emission buildings related to their production, operation and demolition. “ 1 The ZEB includes both research on building features as facade materials, heating systems and research on the end interplay between the end user and energy efficient buildings. This thesis is based on a project connected to the work package 3 of the ZEB-project: “Energy Supply Systems and Services”. The project is cooperation between SINTEF and NTNU. Heat recovery is an important part of the air handling unit in buildings to lower the energy usage due to heating (and cooling) of the supply air. The heat recovery unit does usually include a heat exchanger where the hot exhaust air exchanges heat with the cold supply air (or opposite for cooling purposes). In the Nordic countries flat plate heat exchangers have been the dominating heat recovery system in buildings with several living units and centralised air handling systems. Different from the more energy efficient heat recovery wheels these heat exchangers will not cause the problems of leakages of pollutants from the exhaust to the supply air (Zhang, 2012). However, the aluminium or plastic based flat plate heat exchangers have quite low 1

«About ZEB», Published: 12.08.11, Quoted:12.05.12, http://www.sintef.no/Projectweb/ZEB/AboutZEB/

1

Membrane Based Heat Exchanger

.

annual energy recovery efficiency due to the energy needed for the frost protection systems which are required in low temperature areas (Drivsholm et al., 2005). Condensation and frost problems in flat plate heat exchangers may be avoided if the exhaust air is dehumidified before the temperature reaches the saturation temperature for humid air. In warm and humid areas where cooling and dehumidifying of supply air is required to meet the desired indoor air quality special enthalpy exchangers have been developed. These exchangers utilise membranes to separate the fresh supply air and the exhaust air. The differences in temperature and moisture content on the different sides of the membrane make heat and moisture transfer from the warm and humid side to the cool and dry side by passive transport mechanisms. In a preliminary work by the author the membrane based heat exchanger was found to be able to decrease the frost protection temperature compared to a conventional flat plate heat exchanger. This could lead to great energy savings in Nordic residential buildings with several living units and centralised air handling systems. A method to predict the temperature efficiency and the energy recovery efficiency for a given heat exchanger geometry was derived. The necessary moisture transfer efficiency to avoid freezing problems was also found. The results showed that for the Oslo climate a 70% moisture transfer efficient heat exchanger was needless of a frost protection system and its heat recovery efficiency would reach 90%. This means that if such moisture transfer efficiency is obtainable it can reduce the energy used in frost protection and still the indoor air quality will be improved without increased total costs. This is important for new airtight buildings. (Aarnes, 2011) However, the research on how the flat plate membrane based heat exchanger performs in cold climate conditions is limited. Actually no available research articles were found about the theme. Nevertheless, several manufacturers sell this type of exchangers targetting the cold climate market and claim their good performances. 2 3

2

«Case Study: Natural Resources Canada (NRCan) - Canadian Centre for Housing Technology (CCHT) in Ottawa, Ontario.», Quoted: 12.05.12, http://www.dpoint.ca/energy-recovery-ventilator-cores/casestudyccht.php 3

«Mitsubishi Electric: Canada, Energy Recovery and Ventilation, Benefits.», Quoted: 23.05.12, http://www.mitsubishielectric.ca/en/hvac/erv/benefits.html

2

Chapter 1: Introduction

1.2. Objective Stated above, the membrane based heat and moisture exchanger may have great potential as an energy saver compared to a plastic based heat exchanger. However, this statement was based on the assumption that the membrane will elude the problems concerning condensation and frost formation in the exhaust air channels in the exchanger at low temperatures. Several manufacturers produce membranes that could be suitable to use in a membrane based heat exchanger. Since experimental testing takes a lot of time and effort, a method to predict the moisture transfer effectiveness from available data given by the manufacturers is preferable. Available calculation methods found in the literature includes membrane properties only found through special laboratorial testing. The objectives of the work presented in this thesis are to: 

Develop a method to predict the moisture transfer effectiveness in a membrane based heat exchanger. The input values should not require other properties than the ones usually given in the producers’ datasheets.



Build a test rig to compare plastic based heat exchangers against a membrane based moisture and heat exchanger due to water condensation and frost formation.



Evaluate the membrane based heat exchanger prototype’s appropriateness for heat recovery in cold climate areas.

The mathematical method derived was put into a simple program run in Microsoft Excel. The program should be suitable for further investigation in the suitability of different membranes for use in membrane based heat exchangers. A heat exchanger prototype was developed and a test rig was built in the lab at the Department of Energy and Process Engineering at NTNU to be able to compare different heat and moisture transferring materials. The experimental results will be compared to the results of the mathematical model.

3

Membrane Based Heat Exchanger

.

1.3. Literature The available documentation regarding experimental testing of membrane based heat exchanger performance in cold climates is limited. Nevertheless, there were found two commercial producers of membrane based flat plate heat exchangers designed for cold climate. Both manufactures are operating in North-America. The “Japanese paper total heat exchanger” was invented in 1969 and patented in 1972. The advantages were that this type of heat exchanger could transfer both humidity and heat and without the size and complexity of the energy wheels. (Yoshino & Hashimoto, 1973) These exchangers are still sold under the name Lossnay from Mitsubishi Electric Corporation. The manufacturer claims that this exchanger will work very well without freezing problems in cold climate areas.3 Both the temperature efficiency and the moisture transfer efficiency is claimed to reach 70%. The membranes in this exchanger are micro porous.4 However, no available documented research on the cold weather performance of this exchanger was found. Another commercial membrane based heat exchanger was developed in Canada by dPoint. This firm patented a method to produce a water permeable laminated hydrophilic membrane in 2010 (Huizing, 2010). The dPoint ERV Core was tested in Canadian winter conditions and compared to a standard Coroplast HRV core. The results are only shortly mentioned on the dPoint website and the report is confidential.2 However a very similar experiment was done in summer conditions at the same research centre; the Canadian Centre for Housing Technology and this experiment was published (Ouazia et al., 2006). It is therefore reason to believe that the winter condition experimental tests were performed after the same procedure as in the article of Ouazia (2006). The winter test showed, according to the manufacturer’s website, that the ERV Core had no need of drainage and were 10% more efficient than the standard HRV core. The manufacturer claims that the ERV Core has 80 % temperature efficiency and 40% moisture transfer efficiency. The heat exchanger was working down to a supply air temperature at -19oC without frost protection systems.2 Articles on the sensible and latent effectiveness for different air flow rates(Nasif et al., 2005; Zhang, 2010) temperatures, flow patterns or geometry(Vali et al., 2009; Zhang, 2010) and humidity levels (Kadylak et al., 2009; Niu & Zhang, 2000) were found in the literature. Zhang (2010) used a modified method to predict the moisture transfer effectiveness and compared different heat exchanger geometries by the resulting effectiveness. He found that a quasi-counter flow geometry was both possibly to build and superior to the cross flow geometry regarding effectiveness. He did also perform a CFD analysis to see the flow pattern inside such an exchanger (Zhang, 2010). The result is reprinted below:

4

«Mitsubishi Electric: Canada, Energy Recovery and Ventilation, Features.», Quoted: 23.05.12, http://www.mitsubishielectric.ca/en/hvac/erv/features.html

4

Chapter 1: Introduction

Figure 1.1.

Flow pattern in a quasi-counter heat exchanger. Reprint from (Zhang, 2010).

Condensation will occur at the heat exchanger plate surface if the surface temperature of gets lower than the saturation line for the moist air flowing over it. The saturation temperature for a given absolute humidity, was given as (Aarnes, 2011): (1.1.) (

)

If the surface temperature is below 0oC the condensate will form ice crystals. The surface temperature in the heat exchanger in a given point may be assumed to be: (1.2.) The condensation and frost problems in heat exchangers are strongly linked to the exhaust air humidity. Kalamees et al. (2009) studied the outdoor and indoor humidity and temperatures in 170 Finnish residential buildings over a whole year. The connection between the outdoor temperature and indoor relative humidity may be seen in the reprinted figure below.

Figure 1.2.

The coherence between the outdoor temperature and indoor relative humidity for Finnish residential buildings. Reprint form Kalamees et al. (2009). 5

Membrane Based Heat Exchanger

.

Frost formation on a cold flat plate with moist warm air flowing over it was shown to grow in different forms of different humidity and temperature levels. For quite warm surface temperatures (close to 0oC) and low humidity levels the frost was found to be dense and had a smooth surface. For colder surfaces and high humidities the frost grew as crystals. These ice crystals increased the surface roughness and actually increased the heat transfer coefficient between the plate and the flowing air in the beginning of the experiments. (Yun et al., 2002) The results from the experiments are shown in a reprinted figure below:

Figure 1.3.

Reprint from (Yun et al., 2002). Frost formation on a horizontal cold plate. Heat transfer coefficient dependency on time and distance from leading edge.

The moisture transfer efficiency of the membrane based heat exchanger will depend on the ability of the membrane material to transport water vapour (or the resistance to let the water molecules diffuse through it). Two main models exist to describe permeation of species through a membrane: 1) The pore-flow model 2) The solution-diffusion model The pore flow model describes the transport phenomena when a partial pressure difference at different membrane sides makes a convective driven flow of species from one side to the other through tiny pores in the membrane. The pore size acts like a sieve to sort out the different species.(Mukhopadhyay & Midha, 2008) The solutiondiffusion model was described in a review of Wijmans and Baker (Wijmans & Baker, 1995). Shortly, the model states that a permeant is dissolved (absorbed) in the membrane material and then diffuses through the membrane. The rate of absorption and diffusion is dependent on the concentration gradient. The permeability, P, is then defined as the diffusion multiplied with the solubility: (1.3.)

6

Chapter 1: Introduction

To find the permeability experimental test may be done. To find the solubility gravimetric test is usually made(Modesti et al., 2004). There exist several types of vapour transmission and permeability tests. McCullough (2003) compared 26 different commercial fabrics tested according to five different standard methods. When comparing the water transmission flux for the different tests huge differences were found. Some of the differences can be explained by the differences in test methods; some have large humidity gradients and other smaller. In some test a resistance due to a stagnant air layer appears. Some test let the membranes come in contact with water and others have a great temperature gradient as well as a humidity gradient. (McCullough et al., 2003) This implies that when using available test results to predict the moisture transfer resistance of a membrane an understanding of how the tests was conducted is vital to be able to sort out the internal membrane resistance alone. Henry’s law says that the partial pressure, p, is equal to the concentration of the solute, C, multiplied with a constant . (1.4.) Henry’s law may also be expressed in terms of the volume fraction, φ (as an alternative to the concentration) and a constant k, describing the water molecules affinity to the membrane material: φ

(1.5.)

The relationship between the volume fraction,

and the mass fraction,θ is: (1.6.)

(

)(

)

Where is the molar weight of the membrane material and is the molecule weight of water. Zhang and Niu (2002) used a general sorption isotherm model derived from Henry’s law to describe the relationship between the moisture content in the membrane at different relative humidity levels. The relative humidity is expressed by the partial pressure divided by the vapour pressure at a given temperature. Since the vapour pressure of water at a given temperature is constant this value may be included in the constant k in equation 1.5. and the volume fraction may then be an expression of the relative humidity. This will give the expression given in Zhang and Niu (2002): (1.7.)

Where is a constant, is the relative humidity and is the moisture content in the membrane. is the maximal uptake of water in the material (Zhang & Niu, 2002).

7

Membrane Based Heat Exchanger

.

The water molecules unsymmetrical shape gives the molecules a polarity that makes intermolecular bonds between the molecules possible. This polarity makes also water a good solvent and makes the molecules easily bond to other polar molecules, ions or atomic groups. This feature is utilized in hydrophilic membranes. However, the well developed and described Flory-Huggins theory used to describe sorption of solvents into a membrane may not be used if the bonds between the penetrant molecules are stronger than the bindings between the membrane material and the penetrant.(Favre et al., 1993) The bindings between the water molecules make clusters of water inside the membrane. The approach from Favre et al. (Favre et al., 1993) called the ENSIC model (engaged species induced clustering) is described by looking at the membrane surface as a matrix of either cells of polymer or cells of solvent (water vapour). See picture below:

Figure 1.4.

Water molecules in the gas stream may either bond to polar groups in the polymer membrane surface (white balls) or to other water molecules already absorbed to the surface.

At equilibrium the membrane has absorbed some water. Water vapour molecules may then either be absorbed by polar groups in the membrane material or by other water molecules already absorbed. If the partial pressure rises the number of water molecules absorbed by the membrane may be assumed to rise. If the total number of molecules in the membrane is assumed to be huge compared to the rise in numbers of water molecules due to the partial pressure rise the total molecule number may be assumed constant. This gives the relation(Modesti et al., 2004): (1.8.) The theory may explain why an increase in water content will decrease the moisture transfer resistance of the membrane. The other factor in the permeability equation in 1.3., the diffusivity, is expressed by Fick’s first law (Crank, 1975): (1.9.) F is the rate of mass transfer per unit area and C is the concentration of water vapour. Fick’s second law is expressed as:

8

Chapter 1: Introduction

(1.10.)

Fick’s first law is derived from the free diffusion theory and does not take into account the changes in molecular structures which play an important role in the transportation of water in a hydrophilic membrane. For porous materials the diffusivity may be considered independent form the change in concentration over the membrane. For hydrophilic materials this are not the case and the water vapour diffusion is shown to change a lot with changing humidity levels. (Gibson, 2000) Modesti et al.(2004) showed that the diffusivity changed with different water activities and could by this not be considered to be a constant value. This differed from the model by Zhang (2012) where the water diffusivity in the membrane was assumed constant for a certain membrane.

9

Chapter 2: Mathematical Model

Chapter 2

Mathematical Model In a present work a mathematical model to predict the pressure drop, the temperature effectiveness and the annual heat recovery efficiency for a given frost protection temperature in a flat plate heat exchanger was derived (Aarnes, 2011). The heat transfer in this types of heat exchangers was shown to be independent of the heat exchanger material if the material was sufficiently thin. Hence, the temperature effectiveness is a function of the geometry and the inlet boundary conditions. As written in chapter 1.3. Zhang and Niu (2002) derived a method to find the moisture transfer effectiveness similar to the well-established ε-NTU method for heat transfer. Their method required additional laboratory testing to find the membrane thickness, diffusivity and sorption correlations. In this chapter a method to find the moisture transfer effectiveness without additional testing utilizing the information usually given by the manufacturers is derived. Membranes are often tested in a standardized way. For example the standard ISO 12572: Determination of water vapour transmission. The results are the materials permeability (s/m) given by the mass flux and the difference in the vapour partial pressure over the membrane. In addition it is often possible to find properties as thickness, porosity and water content at given humidity levels (often soaked (100%) or 50% RH).

2.1. Membrane based heat exchanger theory If the heat of evaporation and the density are assumed to be constant the moisture transfer effectiveness may be expressed as (Zhang & Niu, 2002) : ̇

̇

̇

(2.1.)

̇

F is the flow rate of moisture from the exhaust air side to the supply air side (in cold climate). Note that the term efficiency and effectiveness denotes the same feature if the flow rates in each direction are equal. Niu and Zhang (2002) showed that the relationship between the effectiveness and a parameter called the number of transfer units; NTU, for moisture transfer followed the same form as for heat transfer. This means that the existing ε-NTU relations for heat transfer in heat exchangers may also be used for moisture transfer. ε-NTU relations for different exchangers are found in Kays and London (1964) or in Incropera and DeWitt (2007).

11

Membrane Based Heat Exchanger

.

When evaluating moisture transfer with the ε-NTU method the only difference from heat transfer is the definition of the effectiveness given in 2.1. and the NTU-value: (2.2.) ̇

Where is the overall moisture transfer coefficient and ̇ is the flow rate of air through the exchanger. The relation between the effectiveness and the NTU is given as: (2.3.) is the ratio between the minimum and maximum air flow rates. The relations must be found from experimental investigation and different empirical correlations may be found in the literature. The relation is either to be found as empirical equations or as tabulated values. An empirical relation for a cross flow arrangement is found in Kays and London(1964): ( )

(

(2.4.)

)

For a counter flow arrangement the empirical relation under is found in (Incropera & DeWitt, 2007): (2.5.)

(2.6.)

To find the effectiveness for the relevant quasi counter flow arrangement a relation from previous work was used (Aarnes, 2011): )

(2.7.)

For a given geometry and air flow rates the effectiveness may be given from the previous equation if the overall moisture transfer coefficient is found. Different form the heat transfer calculation this value will in the moisture transfer case be dependent on the membrane material as the next chapter will show.

12

Chapter 2: Mathematical Model

2.2. Heat Exchanger Mass Transfer The heat exchanger consists of of multiple membranes that separate the supply and exhaust air.

Figure 2.1.

Flow over membrane.

The transfer of water from the humid exhaust air side to the dry supply air side (for heating applications) consists of mainly two parts: (1. Convection to/from the surface from/into the air steams. This part is given by the advection due to the fluid motion over the surface and the diffusion of water molecules in the air streams. (2. Transport mechanisms inside the membrane material. These two parts may be added in an overall mass transfer coefficient UM: [

]

Where is the mass transfer resistance connected to the convection and the mass transfer resistance inside the membrane.

(2.8.) is

2.2.1. Convective moisture transfer Convection may be divided into two main groups: forced convection and free convection. If the heat convection coefficient h is found, the mass transfer convective coefficient may be found from the mass-heat analogy as described in Incropera and DeWitt (2007): (2.9.)

13

Membrane Based Heat Exchanger

.

Where is the water diffusivity in air, is the thermal conductivity and the Lewis number. is a constant usually set to 1/3. (Incropera & DeWitt, 2007)

is

Forced convection will be the combination of advection due to the bulk motion of fluid and diffusion of water molecules in the air. Free convection is mainly diffusion transport, but the same analogy as for forced convection may be used since the diffusion causes a change in density and by that a buoyancy effect. (Incropera & Dewitt, 2007) In a heat exchanger there will always be a velocity present and the convection of heat and mass will then be a forced type. To find the heat transfer coefficient, a Nusselt number correlation was used. (2.10.)

is the hydraulic diameter. In fully developed laminar flow in a rectangular channel the forced convection Nusselt number is found from(Aarnes, 2011): (2.11.) Where is the aspect ratio (height of channel divided by the width). Using equation 2.10. the heat transfer coefficient h is found and may then be used to find from 2.9. The resistance Rconv in equation 2.1. is then given as: (2.12.)

2.2.2. Moisture Transport through Membranes The moisture transport resistance through a membrane is strongly dependent on the membrane material. There exist different classes of water permeable membrane materials. For use in a moisture transferring heat exchanger either a microporous or a nonporous hydrophilic membrane was found to be most favourable (Aarnes, 2011). The flow rate of water transported through a membrane is easily expressed as the total volume flow rate of air multiplied with the water vapour concentration difference from the inlet and outlet sides of the exchanger channels: ̇

(

)

̇

(2.13.)

Measurement of the water vapour concentration directly is difficult and there is more convenient to express the relation above in terms of the relative humidity difference. 14

Chapter 2: Mathematical Model

The relation between concentration and vapour pressure is given by the perfect gas law: ̅

( )̅

(2.14.)

Where is the molar weight and m is the mass, ̅ is the universal gas constant and is the temperature in Kelvin. The relative humidity is defined as the vapour pressure divided by the saturation pressure at the same temperature. The water vapour concentration is the mass of water in the air divided by the total volume of the air. If the temperature over the membrane is assumed constant 2.13. may be expressed as: ̇

(2.15.) ̅

If the rate of change of concentration through each channel in the heat exchanger is assumed to be constant the moisture flux may be assumed to be given in terms of an arithmetic mean concentration difference and the overall mass transfer resistance in the membrane (Gibson, 2000): ̅

(2.16.)

Since it is more convenient to measure the vapour pressure than the concentrations of water vapour, the permeability constant, P may be introduced: (2.17.) Assuming that the arithmetic mean concentration difference equals the difference over one side combining the equations above gives a relation between the permeability and the resistance. Applying the ideal gas law gives: (2.18.) ̅

With this relation it will be possible to find the resistance and further the overall moisture transfer coefficient from given values for the permeability. Equation 2.18. shows that the resistance is dependent on temperature. However, the temperature dependency was assessed in an article by Gibson (2000). He investigated the temperature dependency of vapour transport through different membrane materials used in waterproof, breathable clothing. He found that the vapour transport resistance inside the membrane changed very little with temperature in the range from 3 - 40oC. This may be explained from that the permeability is shown to increase with increasing temperature (Mondal et al., 2006) due to the increase in free volume radius. Looking at 2.18. the resistance will decrease with increasing temperature if 15

Membrane Based Heat Exchanger

.

the permeability is constant. By using the permeability test temperature value in equation 2.18. the temperature dependency will be eliminated and the resistance may be assumed to be temperature independent since the partial pressure is dependent on temperature.

2.2.3. Humidity Dependent Resistance The water vapour transport resistance was showed to be dependent on the mean relative humidity over the membrane. The relative humidity affected the amount of vapour content inside the membrane. And the water content in the membrane will affect the water vapour transport to a great extent. (Gibson, 2000) Since the relative humidity will vary through the membrane based heat exchanger, a mean moisture transport resistance was used to find the overall moisture transfer coefficient . The normal operation area for such an exchanger in the Nordic climate is displayed in the figure below.

Figure 2.2.

The normal operation area for a membrane based heat exchanger in cold climate. Green lines are relative humidity lines. The normal middle value is from RH approximately 30% to 80%.

It was preferable to find a relation between the resistance and the mean humidity for different types of membranes: (2.19.) Assuming that the water diffusivity in the membrane material was constant it was possibly to see that the correlation from 2.19 follows the same shape for different materials. The relation between the solubility and permeability is given in 1.2. Results from the literature for different membranes are presented below. The sorption curves are displayed in the same form as the resistance assuming 16

.

Chapter 2: Mathematical Model

Figure 2.3.

Upper: Results from Gibson(2000): water vapour transport resistance for different types of membranes for different humidities at 3oC. Lower: Resistance equivalent values from sorption curves from (Marais et al., 2000),(Modesti et al., 2004) and (Niu & Zhang, 2000).

The literature shows that the relation follows the same form for all different materials, both porous and nonporous hydrophilic ones. As possible to see from the figure above the resistance for a soaked membrane (i.e. RH 100%) does not goes to zero, but an asymptotic value. A correlation of this form will give a good approximation of the relation between the humidity level and he resistance: (2.20.) However, to find the three constants A, B and C in this correlation three resistances and corresponding humidity levels need to be known. This may be difficult to find and a correlation with only two constants may be more useful. If a reference resistance is found a relation of this form may be used:

17

Membrane Based Heat Exchanger

.

(2.21.)

Where K is a constant for a given membrane material. The value may be found from a second reference point. (2.22.)

Since only two points are given the resistance outlying from these points would be inaccurate. Especially if the two references are close to each other and the resistance level at interest is at another level than the two. The reference points will give the best result if they are quite far apart in humidity level. However, in most cases the only interesting points is between 30 to 80% RH according to figure 2.2. and there may be assumed that if the first reference point is between 100-60% and the second between 20-50% the results will be good enough for this use. To find the overall moisture transfer coefficient a mean resistance value should be found. The inlet conditions are given in relative humidity. To find the mean humidity level in the exchanger the humidity should be converted to an absolute humidity because this will change linearly through the exchanger different form the relative humidity (see figure 2.2.). (2.23.)

The relation between the absolute and relative humidity is given in Aarnes (2011):

(2.24.)

The mean resistance is then found from: (2.25.)

If a permeability test is run and the permeability is given as in equation 2.17 the reference resistance is given in 2.18. However, permeability test results are not always available, and there is seldom possibly to find more than one reference resistance from permeability tests. This means that there will be preferable to find the resistance from other means as well. For a microporous membrane the resistance is a function of the porosity and the possibility for the water vapour to diffuse through the tiny air filled pores. Since the resistance is dependent on both sorption resistance and diffusion resistance the reference point should be found when the membrane is 18

Chapter 2: Mathematical Model

soaked (RH 100%) then it is possible to assume that the sorption resistance is zero. The may in this case be expressed as:(Yeh & Chang, 2005) (2.26.)

is the diffusion coefficient or water vapour in air, is the porosity, is the tortuosity and t is the thickness of the membrane. The tortuosity may be eliminated by a correlation between the porosity and tortuosity (Alves & Coelhoso, 2004): (2.27.)

For a nonporous membrane the must be found from other relations. All models found to predict the water vapour sorption for different materials build on experimental data. For instance the Guggenheim–Anderson–de Boer model (Huang et al., 2004) includes three different constants found from experimental investigation. Another model used in calculations for a membrane based heat exchanger was derived by Niu and Zhang (2000): (

(2.28.)

)

The relation between the water content in the membrane and humidity needs to be known to find the sorption constant C. is the maximum amount of water the membrane will hold(kg water/ kg membrane). is the diffusivity of water in the membrane. If a reference value is found: (

)

(2.29.)

(2.30.)

If a sorption test at a second reference humidity is done the resistance for that humidity levels if found from:

19

Membrane Based Heat Exchanger

.

(2.31.) (

) (

)

2.2.4. Evaluation of the empirical relation The approach derived in the previous chapter used and compared to the results form Gibson (2000). He tested several membranes for the dependency on resistance on relative humidity. See figure 2.3. McCullough et al. (2003) tested several fabrics after different permeability testing standards. Two of them were Sympatex® from Akzo-Nobel and Gore-tex® from W.L. Gore & Associates. The first is a nonporous hydrophilic membrane and the latter is micro porous. The vapour flux found from ASTM E 96 B (ASTM, 2011) and JIS L 1099 (JSA, 1999) in the article by McCullough et al (2003) were used to find two reference resistances at 75% and 100% (soaked membrane) relative humidity respectively. The resistance in the air layer in the ASTM E 96 B-method was approximated after the relation: (2.32.)

Where is the height of the air layer, is the diffusivity of water vapour in air and is the test cup opening area. The comparison between the proposed method and the result from Gibson is shown below:

Figure 2.4.

Comparison between measured values by Gibson(2000) and empirical relations by the author.

20

Chapter 2: Mathematical Model

The model seems to fit quite well for humidities over 40% with the results from Gibson (2000). The porosity of the Nafion®117 membrane from DuPont was showed to be 52±5% (Chen et al., 2008) and was used to find a reference resistance at 100% RH from 2.27. A sorption curve for different humidities was found in (Peron et al., 2010). Values at 20 and 100% RH were used to find a resistance form the sorption model (2.31.).

Figure 2.5.

Comparison between measured values by Gibson (2000) and empirical relations by the author.

The model does also in this case fit very well with the results from Gibson (2000) for humidity levels between 40 and 60%.

21

Membrane Based Heat Exchanger

.

2.3. Practical use of the mathematical model For practical use of the model a simple program written in Visual basic language called HXcalc was written. The software used was Microsoft Excel 2010. This is an extension to the simple program HXOpt made in previous work (Aarnes, 2011). The program code is to be found in appendix A.3. The user interface has many dropdown choices for: changing parameter, geometry, membrane resistance calculation method, energy effectiveness weather data location etc. A print screen picture of the input interface is showed here:

Figure 2.6.

User interface of the input sheet in the developed HXCalc program. The white boxes are to be filled in by the user.

After clicking the “Run simulation!” button the temperature effectiveness, moisture transfer effectiveness (if “Membrane HX?” is set to “Yes”) and the annual heat recovery efficiency will be displayed in a separate worksheet named “Results” as graphs where the effectiveness is the y-values and the chosen “Parameter to change” as the x-axis. The results will appear as points. Clicking on a single point will make a 22

Chapter 2: Mathematical Model

separate window appear. The window includes a drawing of the exchanger with additional information from the conditions given in that point. See example below:

Figure 2.7.

Results sheet after clicking on a single point. An information window appears.

In addition will a second result sheet called “Hum-temp” update when clicking on point in the “Result” sheet. In this chart the input and output temperature and absolute humidities will appear as lines (between inlet and outlet values) in a temperature-humidity diagram. These lines indicate the temperature and humidity development through the exchanger. See picture below for the same case as the picture above:

Figure 2.8.

Results sheet after clicking on a single point. An information window appears.

23

Membrane Based Heat Exchanger

.

2.4. Evaluation of the Method in HX Applications The moisture transfer effectiveness calculation method derived in the previous chapters was compared to the results in an article by (Kadylak et al., 2009). In this case only the moisture transfer effectiveness was addressed hence the article addressed the effectiveness of a humidifier operating in a proton exchange membrane fuel cell. Kadylac et al.(2009) took their starting point in Zhang and Niu’s(2002) method to calculate the moisture transfer effectiveness in a flat plate heat and moisture transferring heat exchanger. Two reference points for the mean resistance values were found from the relation in 2.28. The C value was set to 10. Then the mean resistance was found from equation 2.25. The results comparing to both Kadylac et al. (2009) and Niu and Zhang’s (2000) results are displayed below: Simulation input values: LC

0.5m

H

0.145m

Nr of plates

29

̇ C

0.039 m3/s 10 70.15

o

C

0.23 kg/kg 100% Le

1.2 0.0263 w/mK 1.17 kg/m3 1.33*10-5 m2/s

1010 J/kgK

Figure 2.9.

Comparison between the moisture transfer effectiveness calculations methods on changing supply inlet relative humidity from Aarnes, Kadylac et al.(2009) and Niu and Zhang(2000).

As seen the derive method has the same evolution as the results from Niu and Zhang (2000), but the results are a bit lower value. At high humidities the method correlates well with Kadylack at al.’s ( method). Second the dependency on the flow rate was compared to the article by Niu and Zhang (2000). The results are shown below:

24

Chapter 2: Mathematical Model

Figure 2.10.

Comparison between the derived model and results from Niu and Zhang (2000) for changing flow rate.

The results above have the same input values at the previous case. The proposed model correlates well with the model of Niu and Zhang (2000) for this specific example. The temperature effectiveness is not dependent on temperature, different from the moisture transfer effectiveness. A comparison was done, still with the same inputs as over, for different supply inlet temperatures and a supply inlet humidity kept constant at 50%. The results were again compared to the result in the article by Niu and Zhang (2000):

Figure 2.11.

Comparison between the derived model and results from Niu and Zhang (2000) for changing supply inlet temperatue.

25

Membrane Based Heat Exchanger

.

The models do not correlate well regarding the moisture transfer effectiveness dependence on temperature. As seen is the proposed model less sensitive to changes in supply inlet temperatures than the model by Niu and Zhang (2000). The reason is probably that the derived model uses a mean value for the moisture transfer resistance while Niu and Zhangs’ (2000) model is iterative and may then take into account differences in humidity through the exchanger. The annual heat recovery efficiency from the method derived in the previous work by the author (Aarnes, 2011) was included in the program HXcalc. Zhang and Niu (2001) derived a method to predict the annual energy savings using a membrane based heat exchanger. A simulation with the same inputs as used in the article by Zhang and Niu (2001) was done. The weather data used was from Hong Kong and was found at the website of the American Energy Department1. Under are the results compared when only looking at energy for heating. The two methods seem to correlate very well. The author’s model gave a bit higher resulting values; this is probably because the model from Zhang and Niu (2001) did not take into account heating in the summer months as the author’s model.

Figure 2.12.

Required and saved energy for heating. Comparison between the derived model and results from Niu and Zhang (2000).

1

«Department of Energy: EnergyPlus Energy Simulation Software: Weather Data: China.», Quoted: 06.06.12, http://apps1.eere.energy.gov/buildings/energyplus/cfm/weather_data3.cfm/region=2_asia_wmo_region _2/country=CHN/cname=China

26

Chapter 3: Experimental Investigation: Method

Chapter 3

Experimental Investigation: Method The objective of the experimental part of this work was to compare a plastic based heat exchanger to a membrane based heat exchanger due to the problems of condensation and freezing in cold climate conditions. The fundamental hypothesis for the development of the test method was: When ice and frost is formed inside the

exhaust air channels in a heat exchanger the pressure drop over exchanger on the exhaust air side will rice. The pressure drop on the supply air side will remain constant through the test period. In addition the test rig was used to perform experiments that could be compared to the mathematical method derived in the previous chapter.

3.1. Development of the Test Rig A test rig was built to compare the frost formation in two plastic sheets and a membrane based heat exchanger. The test conditions should be as stable as possible over the test periods and it was preferable to have quite large differences in both humidity and temperature level in the supply air and exhaust air sides respectively.

3.1.1. Supply Air Side The supply air was taken from the lab and blown through the cooling coils with a 12V axial computer fan. The air flow rate was regulated by covering the inlet area to the fan with air tight tape. Two cooling coils in series with glycol as the chilled fluid were used to cool down air on the supply air side. The initial testing of the cooling coils is described in appendix A.1. The glycol was cooled down to -25oC. The long distance between the glycol compressor and the cooling coil gave an immense temperature loss and the coldest outlet air temperature measured at the outlet of the cooling coil was about -15oC. The lack of space in the lab made it necessary to build about 2.5m of ducts from the cooling coil to the heat exchanger. Plastic ducts of 3cm diameter were used and isolated with Armaflex (synthetic rubber isolation for ducts) from Glava. Due to the very low velocity inside the ducts it was impossible to insulate the ducts well enough to conserve the cold temperature. When reaching the heat exchanger the temperature had riced to about -4 to -10oC depending on the air flow rate. The humidity and temperature probes were placed in square shaped ducts made out of stainless steel plate. The ducts worked also as air straighteners before and after the heat exchanger, supplying the air flow as even distributed as possible. 27

Membrane Based Heat Exchanger

.

One centimetre from the channel opening, close to the heat exchanger a thin pipe was welded to the channel and a small hole was drilled through the channel wall. This was used to measure the static pressure over the heat exchanger. Due to ice formation inside the cooling coil the air flow rates decreased after approximately 12 hours. The tests were therefore ended after 12 hours.

3.1.2. Exhaust Air Side Heating and moisturising of air to simulate the exhaust air side was done inside a “climate chamber” box built of 40mm thick polystyrene plates. By taking some of the cold air from the cooling coil the inlet air to the “climate-chamber” was stable regarding temperature and humidity. A bucket of water with a 300W heating element and a thermocouple coupled to a PID regulator was used to humidify the air. By regulating the water temperature in the water bucket both temperature and humidity were controlled since the inlet temperature and humidity were practically constant. See the picture below.

Figure 3.1.

Drawing made in Google Sketch Up. The exhaust air was heated and humidified in a polystyrene box placed under a table to make the rig as compact as possible. The duct to the lower right was connected to the cooling coil.

A dimmer was connected to the heating element to regulate the temperature of the water. To get a stable temperature and humidity levels it was preferable to have the same effect from the heating element at all time. The heater was dimmed to give a water temperature at 34oC. This gave an exhaust temperature at approximately 22 oC and humidity at approximately 42% RH. The PID-regulator had a set point at 40 oC to protect the bucket from dry-out. A small axial computer fan with 12V power supply was attached at the inlet of the exhaust air supply duct. This was made out of a 10cm diameter spiro-channel insulated with an aluminium foil covered fibre glass insulation mat from Glava. The spiro-channel was attached to a rectangular duct

28

Chapter 3: Experimental Investigation: Method

described in the previous chapter. An identical duct was placed in the opposite side of the heat exchanger.

3.1.3. Small Scale Heat Exchanger Prototype The heat exchanger was build up with a frame made of two hexagonal shaped 6 mm transparent acrylic plastic plates. Two 3mm bends cut out from a Lexan plate were glued to each hexagonal plate using epoxy glue. Four more bends of 6mm Lexan were also cut out. The first membrane layer was connected to the frame base layer using double sided tape. The construction was built like a sandwich with Lexan bends and membranes as seen in the picture below. For the first test this prototype was used. However, for the rest of the experiments eight new bends, now of 3mm Lexan were made to increase the membrane layers from 3 to 5. The mechanical drawing of the prototype frame is to be found in Appendix A.5.

Figure 3.2.

Heat exchanger made of sandwich construction. Drawing of concept in Google Sketch Up. 3 layer prototype.

Three different types of heat exchanger plate materials were tested. Two nonpermeable “plastics” and one membrane with high permeability to water vapour. Properties of the materials are shown below:

Table 3.1.

Properties for the different tested plate materials.

Material

Manufacturer

Water permeable

Elastic

Crumples in high humidity

Transparent overhead sheets PP Membrane X

Optimax

No

No

No

DuPont DuPont

No Yes

Yes Yes

No Yes

The transparent overhead sheet from Optimax was non-elastic and quite stiff. This made it easy to construct the heat exchanger prototype. The PP-sheets and membrane from DuPont were much more difficult to handle in the construction of the prototype exchanger. Results from permeability tests for the membrane were given in datasheets from the manufacturer. These values will not be given in this report due to 29

Membrane Based Heat Exchanger

.

confidentiality, neither will the actual name of the membrane (therefore is it called X). The PP-sheets were actually the protection layer for the membrane. This material was used since the appearance was very similar to the membrane. The material was probably polypropylene since it was written PP on the sheets. The heat exchanger was placed in a die made from the four rectangular ducts and made air tight with a sealing compound. An isolation mat was placed on top of the heat exchanger under the experiments to minimize the temperature loss to the surroundings. See picture below:

Figure 3.3.

Die for placing of the heat exchanger. The rectangular ducts with pressure measurement pipes are shown.

3.1.4. HSE A HSE report was written before the experiments were run. The whole report is to be found in Appendix A.6. The biggest risk was connected to the use of the ethylene glycol loop used to chill the supply air (leakages, pipe rupture etc.). The risks were found to be acceptable to run the experiments.

3.2. Measurement and Instruments To measure temperature and humidity four instruments from Vaisala were used. Three of them were of the type HMP 233 and the last was of the type HMT 334. The probes could measure both humidity and temperature. They were mounted perpendicular to the air streams in each of the rectangular ducts receiving and supplying air from and to the heat exchanger. The pressure drop over the heat exchanger was measured with two micro manometers from DPM. The relative humidity and temperature of the surroundings and the supply air outlet velocity were logged with the TSI Velocicalc 9555-P instrument. The accuracy of the instruments is shown in the table below: 30

Chapter 3: Experimental Investigation: Method

Table 3.2.

Uncertainty of measurement instruments used in experimental set up.

Manufacturer Model name Measurement

Instrument Uncertainty

Vaisala

HMP 233

Relative humidity Temperature

±2% (3% for 90-100%RH) ±0.1oC

Vaisala

HMT 334

Relative humidity Temperature

DPM

TT470s

Pressure drop

TSI

Velocicalc 9555-P

Relative humidity Temperature Velocity

±(1+0.008*reading) %RH ±0.2oC (at 20oC), ±0.3oC (at 10oC) ±1% of reading ± 0.1Pa +analogue output: ±0.3% ±3% ±0.3oC ±max(3% of reading, 0.015m/s)

The humidity probes were calibrated with a LiCl-salt solution and against a reference value given by the TSI-instrument in a high humidity chamber (the polystyrene box with the humidity and heating bucket). The most important reason for this calibration was to make sure that the four different humidity probes were calibrated against each other, so that the difference between the inlet and outlet humidity level was as accurate as possible. The instruments were not calibrated for temperature measurements since this accuracy was good in the first place. The flow rates were not possible to measure with any availiable instruments due to the very low air flow rates. Since velocity measurement by traversing over the ducts with a velocity measurement instrument was found to be to inaccurate a manual method of measuring the air flow rates was used: A plastic bag was hold tight over the air outlet for 10-20 seconds measured with a stop watch. The bag was then quickly removed and the opening twisted. The bag was then put into a bucket of water to find the air volume. The measurements were made several times to find an average value. The humidity and temperature probes and the two micro manometers were connected to a computer via a serial bus. The program LabView was used to transform the analogue signals to preferable variables and log the results. The pressure drop, humidity and temperature were logged every second. The surrounding humidity and temperature were logged by the Velocicalc every 10 minutes (averaging by the instrument). Microsoft Excel was used to process the outputs into readable charts and averages. The measurement set up is shown in the figures below and at the next page:

Figure 3.4.

Measurement set up.

31

Membrane Based Heat Exchanger

Figure 3.5.

.

Flow chart of the experimental set up.

32

Chapter 3: Experimental Investigation: Method

3.3. Uncertainty The uncertainty of the experimental investigation was connected to the reliability and the accuracy of the measurement instruments and the use of them. The relevant uncertainty can be divided into two separate parts(Novakovic et al., 2007): 1) Random uncertainty. Instrument accuracy, influence from other sources etc. 2) Systematic uncertainty- measurement calibration and upset of instruments etc. An example of random uncertainty is shown in the picture below. The picture shows the pressure drop results, a part of the LabView interface during an experiment. The peak shows random uncertainty.

Figure 3.6.

Random uncertainty in the pressure drop measurement. Print screen from LabView Front Panel interface.

The standard deviation tells something about how the measurements will vary about the “true” value. The standard deviation may then be found from: ( √ (

̅) )

(3.1.)

Where ̅ is the mean value and n is the number of measurements. Since the pressure drops, especially on the exhaust air side, changed trough the test period (in some experiments) using the variation about a mean value for the pressure drop would lead to a to larger standard deviation for the random uncertainty than it actually was. Therefor was this relation used instead of 3.1.: ( √ (

̃) )

(3.2.)

Where ̃ is the “true” pressure drop given in a certain time step by a linear trend line describing the increase in pressure drop with time. This trend line is found by the 33

Membrane Based Heat Exchanger

.

trend line tool in Excel. The supply air and exhaust air side pressure drops are shown in the figure below.

Figure 3.7.

Pressure drop measurement. A linear trend line equation for exhaust air pressure drop is displayed to the right.

The pressure drop was logged every second. If the hole time period is taken into account:  

Exhaust air side pressure drop standard deviation: 0.0621 Pa Supply air side pressure drop standard deviation : 0.2228 Pa

Since the random uncertainty seemed to increase a great deal after 900 min it may only be interesting to look at this first part. If the standard deviation only takes into account the first 900 min:  

Exhaust air side pressure drop standard deviation: 0.0536 Supply air side pressure drop standard deviation: 0.0637

It appeared that the random uncertainty was less for the exhaust air measurement than the supply air measurement. This may be explained by that when the battery capacity was low the analogue output signals from the micro manometers were greater than if the battery was full. This seemed as a reasonable cause seen that the random uncertainty increased with time (ref. figure 3.7.). Random uncertainty for the mean value is calculated as:

√

(

̅) )

(

(3.3.)

The systematic uncertainty may consist of different parts. The first is the one that is given by the manufacturer. This uncertainty is given in table 3.2. The systematic uncertainty for the resulting parameters as the efficiency will also be given by the difference in the calibration of the different humidity and temperature probes. This is possibly to see in this picture: 34

Chapter 3: Experimental Investigation: Method

Figure 3.8.

Humidity level measurement. Chart from LabView interface. The graph shows the humidity level at the same sport for the four different Vaisala probes. The HMP3 probe gives a lower value of the humidity than the other probes, about 0.3%.

The total uncertainty is then: (3.4.)

√

Since the measurements were logged every second and the experiments were run for more than 12 hours the random uncertainty was ignorable compared to the systematic uncertainty for temperature, humidity and pressure as shown in appendix A.4. For the air flow rate measurement, which was manually done by a plastic bag, a stop watch and a bucket of water, the uncertainty was of the combined type. (3.5.) ̇

Since the volume was manually found from between litre marks on the buckets the systematic uncertainty from manually readings may be set to ±0.2 l. The systematic uncertainty from the time to connect and release the bag from the duct outlet and the start and stop of the stopwatch may be set to ±0.5 sek.

Table 3.3.

Uncertainty of parameters in manual flow measurement method. Numbers from experiment 1.

Time (s) 10.17 9.88 9.88 10.12

̅

Volume (l) 4.6 4.3 4.4 4.3

35

̅̅̅

Membrane Based Heat Exchanger

.

The uncertainty of an assembled result parameter is given from (Novakovic et al., 2007) : (3.6.) √(

)

(

)

For the flow rate measurement:

√( ̇

(3.7.)

̅

)

̅

(

⁄

) ̅ ⁄

The flow rate for the example in table 3.3. is then 1.58 ± 0.11 m3/h. The relation between the absolute humidity (used to find the moisture transfer efficiency) and the relative humidity was derived for 0oC in the project report by the author. (Aarnes, 2011) : (3.8.)

(3.9.) √(

)

(

)

(3.10.) √(

)

(

)

For the temperature efficiency the most accurate results were found when the temperature difference at the exhaust air side was used in the temperature efficiency relation. The reason was that the temperature loss to the surroundings was greater on the supply air side than on the exhaust air side. For the moisture transfer efficiency it was assumed to be no loss to the surroundings. The efficiencies for moisture transfer and heat transfer for the exhaust air side are given as: (3.11.)

36

Chapter 3: Experimental Investigation: Method

(3.12.)

The uncertainties are found from: (3.13.) (

√

(

)

)

( (

(

)

) )

(3.14.) (

√ Where

(

) ( (

)

(

)

) )

is another term for the total uncertainty of the given parameter.

37

Chapter 4: Results

Chapter 4

Results In this chapter the results from the mathematical simulations and the experimental investigation will be briefly presented. All results from each experimental investigation may be seen in appendix A.4. Comparison and discussion will be presented in the next chapter. The experimental results are given as temperature and moisture transfer efficiency and not effectiveness since the flow rate measurements were too inaccurate and incomplete. The connexion between the moisture transfer efficiency and effectiveness is shown here: ̇

(

(

̇ ̇ ̇

)

)

(4.1.)

(4.2.)

The relations will be equivalent for the temperature effectiveness/efficiency with replacing with .

4.1. Results from the Mathematical Model To compare the experimental data with the mathematical model derived in chapter 2 simulations were run in the HXcalc program. The input values and results for one run are shown below and on the next page:

Table 4.1. Simulation input values. LC

0.1264m

1.33*10-5 m2/s

L

0.107m

1006 J/kgK

W

0.178m

206.5 s/m

H

0.024m

39%

Number of plates

5

76.27s/m

0.8

100%

0.025 w/mK

45% 3

1.17 kg/m

39

Membrane Based Heat Exchanger

Figure 4.1.

.

Temperature, moisture transfer and annual heat recovery effectiveness for a set point temperature at 18 oC for different air flow rates(m3/s).

The same geometry (see Appendix A.5) and input values were used in the simulation as in the experiments. The input properties were -6.4oC and 33.4% RH for supply air and 22.3oC and 43% RH for exhaust air for the case displayed above. These values were the mean input values from all experiments. Two moisture transfer reference resistances for the DuPont membrane X were found from the permeability test results given in the datasheet from the manufacturer. The reference mean humidities were 39 and 100% respectively from the ASTM E 96B and BW test methods (ASTM, 2011). The air layer resistance for test B was assumed to be 500s/m as in chapter 2.2.4. For flow rates in both directions at 1 m3/h the effectiveness dependence on supply inlet temperatures (all at 33.4% RH) are shown below:

Figure 4.2.

Simulated supply inlet temperature dependence for the efficiencies. Square shapes are the moisture transfer effectiveness and diamonds are temperature effectiveness. 40

Chapter 4: Results

The figure above was taken directly from the HXcalc program. When clicking on the point for -10oC this picture appears on the screen:

Figure 4.3.

Popup window from HXcalc for supply air temperature at -10oC.

4.2. Flow Pattern inside the Heat Exchanger A heat exchanger with coloured copy paper as heat transfer plates was built to see the flow pattern inside the heat exchanger using a “smoke pen” form Björnax AB. A fan was used to pull air through the heat exchanger and the smoke pen was ignited and moved along the inlet end of the heat exchanger. The coloured paper made the smoke visible and it is possible to identify stream lines inside the heat exchanger. Three different picture series were taken starting with the first one with dry paper. After the first series the heat exchanger was exposed to mist form the humidifier for approximately 5 min and a second picture series was taken. After approximately 10 more minutes exposed to mist a third photo series was taken. The paper started to crumple while exposed to humidity and for the last picture series this was clearly visible. yyys fdgfg

Figure 4.4.

Middle streamline for three different photo series. Picture to the right shows a streamline when the paper was exposed to high humidity.

41

Membrane Based Heat Exchanger

Figure 4.5.

.

Adobe Photoshop was used to overlay several pictures to see the different streamlines. Dry heat exchanger to the left, exposed to humid air in the middle and wet heat exchanger to the right.

When the heat exchanger was dry it was possible to see that the air flows over most of the heat exchange area. When the paper started to be wet the flow covered less of the area. This means that there may have been stagnant air in some areas in the heat exchanger.

4.3. Results from the Experimental Investigation The results from the different experiments are assembled for easier comparison in this chapter. For more results from the individual experiments see Appendix 4. For the experiments the resulting parameter is sown as efficiency given from the exhaust air side, not as the effectiveness given from the HXcalc program. Below is a table with the mean results and the name of the heat and moisture transfer plate material for eight different experiments.

Table 4.2. Overview of all experiments with mean values for inlet temperatures and relative humidity, pressure drops, measured air flow rates and calculated efficiencies based on the exhaust air side. ̇

Plate material

̇ ⁄

⁄

1

Optimax

-5.27 23.81 27.4

43.6

2.40

2.55

1.58*

1.58

0.27

2

Optimax

-8.05 20.85 33.6

39.3

3.95

5.79

1.66

1.05

0.37

3

DuPont X

-4.96 21.25 27.1

42.86 9.248 8.96

0.74

1.38

0.41 0.37

4

DuPont PP -8.41 21.04 35.17 46.15 6.16

6.73

1.38

1.30

0.35

5

DuPont X

-0.23 22.91 39.02 45.25 10.47 9.85

1.53

1.33

0.54 0.49

6

DuPont X

-4.32 22.77 29.54 43.27 11.13 10.48 1.55

1.20

0.54 0.58

7

DuPont X

-10.5 23.21 41.04 37.27 27.19 25.60 2.6

0.6**

0.60 0.91

8

DuPont X

-9.62 22.90 34.22 46.60 25.25 24.80 1.4*

0.6**

0.61 0.88

*)

Flow rates not measured, but assumed from velocity and humidity-temperature diagram lines.

42

Chapter 4: Results

The temperature efficiencies for the different experiments are shown below. The temperature efficiencies are quite stable over time for all experiments except experiment two where the temperature efficiency first rice and then sinks.

Figure 4.6.

Temperature efficiency for all experiments.

The development of the pressure drops are shown in the figure below for the experiment one to six. The two last experiments had much higher pressure drop and are therefore displayed in a separate figure:

Figure 4.7.

Pressure drop for all experiments except 7 and 8.Experiement 1,2 and 4 were plastic based while 3,5,6,7 and 8 were membrane based.

43

Membrane Based Heat Exchanger

Figure 4.8.

.

Pressure drop for experiment 7 and 8.

The start pressure drops changed quite much between the different experiments as possible to see in the figures above. However, a significant difference between the plastic based heat exchangers and the membrane based one is to be easily recognised. While the difference in pressure drop between the two air steams was quite constant over time for the membrane based heat exchanger in experiment 3,5,6 and 7 the difference increased quite significantly with time for the plastic based heat exchanger prototypes in experiment 1,2 and 4. A tiny decrease in the pressure drop difference happened in experiment 8. Since the supply air pressure drops were not stable as expected an equation to elude the changes in supply air pressure drop to see the changes at the exhaust air side was set up: ̅

(

)

(

)

(4.3)

Below is a table showing the results for the eight experiments using the equation above for hour based averages in hour 2 and 14 (hour 1 and 11 in experiment 6, and 1 and 6 from experiment 7):

Table 4.3.

Difference in pressure drop over time. Experiment 1 2 3 4 5 6 7 8

̅

̅

0.027497 0.25829 0.008796 0.093076 0.003044 0.014223 0.043156 0.021711

44

0.934894 8.781845 0.299074 3.16458 0.103503 0.483596 1.467318 0.738186

Chapter 4: Results

The membrane based experiment 2 and 4 stand out with high values. High value was also found for the membrane based experiment 7, but this experiment was only run for six hours. The moisture transfer efficiency results from the membrane based heat exchanger experiments are displayed here:

Figure 4.9.

The moisture transfer efficiency dependency on time for the different experiments with membranes.

The results above divide into two groups with experiment 7 and 8 having much higher efficiencies than the other three experiments. This should be seen in connection to the ratio between the flow rates which are much greater in experiment 7 and 8 than in experiment 3,5 and 6. The moisture transfer efficiency dependence on supply inlet temperature:

Figure 4.10.

The moisture transfer efficiency on the exhaust air side dependence on time for the different experiments with membranes.

45

Membrane Based Heat Exchanger

.

The correlation between the temperature and moisture transfer efficiency in experiment 5,6,7 and 8 forms a increasing efficiency trend for lower temperatures. Experiment 3 does not follow this. However the experiment three was the only experiment where the exhaust air flow rate was greater than the supply air flow rate. Ice and condensed water were found in all plastic exchangers (experiment 1, 2 and 4) plus in the last membrane based experiment (experiment 8). The ice formed in different areas in the plastic based and membrane based exchanger respectively. For the plastic based heat exchangers ice formation occurred in the exhaust air channels near the supply air inlet. In the membrane based heat exchanger the ice formed near the supply air outlet. In the membrane based exchanger the membranes tend to expand and crumple in very humid conditions. This happened in experiment 8 and it was also observed when the cooling coil was turned off between the experiments. However, the membranes tightened when the humidity level went down again.

Figure 4.11.

Ice formation in experiment (from left) 1,4 and 8.

4.4. Expansion of the Membrane in High Humidity Conditions A very simple experiment was done with the membrane based heat exchanger. The exchanger was dipped in water for just a second. The membranes expanded and the membranes stuck together as showed in the picture below. After several hours the membranes dried and were stretched out again as before. Below are two pictures showing the difference between the dry and the wet heat exchanger:

46

Chapter 4: Results

Figure 4.12.

Left: Dry heat exchanger. Right: Wet heat exchanger

47

Chapter 5: Discussion

Chapter 5

Discussion The first objective of this work was to make a direct mathematical calculation method to predict the moisture transfer effectiveness. The simulated results and the experimental investigation were compared to see if the proposed method gave reliable results. The second objective was to investigate the difference between the different heat exchanger plate materials regarding condensation and freezing. The change in pressure drop over the heat exchanger was used as an indicator. The heat exchanger prototype top and bottom were made out of transparent acrylic plates. This made it possible to do a visual investigation on if and where ice formation took place.

5.1. Comparison between the Mathematical Model and the Experimental Investigation A comparison between the proposed mathematical method and the method derived by Niu and Zhang (2000) to predict the heat and moisture transfer efficiency was done in chapter 2.4. The comparison showed that the two methods correlated very well regarding changes in flow rate. However, the proposed method by the author did not correlate with Niu and Zhangs’ (2000) method regarding changes in supply temperature. The HXcalc-program was run with the same inputs (inlet temperatures, humidity, flow rates and geometry) from all experiments (see Appendix A.4). A comparison between the simulated and experimental results for temperature efficiency and moisture transfer efficiency are shown in figures on the next page. Since the results from the simulations were given as effectiveness, while the experimental results were given in efficiency it was necessary with a calculation between the two properties. The correlations are shown in equation 4.1. and 4.2.

49

Membrane Based Heat Exchanger

Figure 5.1.

.

Comparison between the experimental and simulated results for temperature efficiency (left) and moisture transfer efficiency (right). Trend lines equations for the correlation is displayed inside the charts.

For both temperature efficiency and moisture transfer efficiency the correlation of the two set of results is close to a trend line. However, the trend line for temperature efficiency does not have the 1:1 correlation as expected. One explanation why the simulated values for the temperature efficiency were higher than the measured ones may be that the temperature loss to the surroundings from the point where the supply air inlet temperature was measured to the heat exchanger was huge. The supply air inlet temperature was included in the definition of the temperature efficiency (equation 4.1. and 4.2.) and if this measured value was lower than the actual inlet temperature to the heat exchanger the temperature efficiency would be calculated to be too low. The points below the trend line in the figure to the left above were for the lowest supply air inlet temperatures which confirms the theory about temperature loss to the surroundings since the colder the supply air temperature, the higher the loss. The moisture transfer efficiency correlation differs with only 10% between the simulated and experimental results. This is really promising since the mathematical method only takes into account a mean moisture transfer resistance value to calculate the effectiveness. This differs from Niu and Zhangs’ (2002) method to calculate the moisture transfer effectiveness which needs a iterative solution method. This result implies that the information given by the manufacturer regarding water vapour permeability was enough information to predict the moisture transfer effectiveness for the given membrane. Nevertheless, as mentioned before, the permeability tests are done in many different ways as stated in the article by McCollough(2003). An understanding on how the tests were performed is vital for good results. To see if the shape of the heat exchanger worked as intended (counter flow with cross flow headers) the streamlines in the dry exchanger were assumed to be of the same form in both exhaust and supply air streams. The picture below shows the streamlines for both streams overlaid.

50

Chapter 5: Discussion

Figure 5.2.

Idealistic quasi-counter flow heat exchanger flow to the left and the results from the smoke-pen test to the right.

This small experiment showed that in the cross flow headers the flows were fairly perpendicular, but in the counter flow area the streams were not counter wise but approximately 30o. This means that to predict results as counter flow will probably give over predicted results as for the temperature efficiency in figure 5.1. Compared to Zhang’s (2010) CFD-model the streamlines in this experiment were less counter flow angled. However, the geometric shape of Zhang’s (2010) exchanger was slimmer than the tested heat exchanger. This may have caused the differences.

5.2. Pressure Drop and Flow Rates through the Exchanger The HXcalc program calculates the pressure drop with this correlation (Aarnes, 2011): (

)

(5.1.)

An increase in the flow rate will then cause an increase in the pressure drop. Increasing flow rates will also cause a decrease in the U-value for the heat exchanger. Regarding to the mathematical method this will decrease the efficiency as seen in figure 4.1. This means that an increase in pressure drop should mean a decrease in temperature efficiency. However, the experimental results show the opposite result. There was not found any correlation between the measured pressure drop over the exchanger and the flow rate as seen in the figure below:

51

Membrane Based Heat Exchanger

.

Figure 5.3. Correlation between the temperature efficiency and the mean pressure drop.

Figure 5.4. Correlation between the exhaust flow rate and the exhaust side pressure drop (left). Correlation between the supply flow rate and the supply side pressure drop (right). Error bars represent the total calculated error. When only one flow rate was measured the error was set to 0.26 m3/h which was the maximum calculated flow rate error for all tests. This result may indicate other reasons different from the flow rate that caused the pressure drop. The pressure drop over the supply air side was not stable in all experiments. In experiment 2,3,7 and 8 the supply air side pressure drop decreased, while in experiment 4,5 and 6 it increased. In both experiment 2 and 4 a great amount of ice formation was observed. The test conditions were quite similar in these two experiments. However the pressure drop over the supply air side had opposite developments in these tests. See figure 4.7. A possible reason for the difference may be differences in the material properties. The experiment 2 utilized Optimax’s quite stiff non-elastic transparent plastic sheets while experiment 4 was utilizing elastic PPsheets form DuPont. While an ice layer was building up in the PP-exchanger blocking the exhaust air channels and increasing the exhaust air side pressure drop, the elastic material changed the channel height of the supply air side channels as well, increasing the pressure drop also here. This did not happen in experiment 2 using stiff plastic sheets. When the air flow at one side was greater than the other the membrane tend to expand and making the channels with the lowest air flow rate narrower. This caused that the pressure drop at this side increased as well and that the pressure drops were almost equal at both sides. If the side with the highest flow rate decreased the flow rate the other side followed the tendency and the pressure 52

Chapter 5: Discussion

drop on each side decreased. This may be seen by comparing the figures 4.7. and 4.8. The only experiment that did not follow this was experiment two there inelastic transparent plastic sheets were used. This effect is really a drawback for the membrane from DuPont in this application.

5.3. Evaluation of the Test Rig For simplicity when building the experimental set up, the already installed glycol cooling loop in the lab was used to produce cold air. Since the air flow rates were kept at low levels to increase the temperature efficiency the temperature loss to the surroundings through the distance from the cooling coil to the heat exchanger was huge. This restricted the cold side temperature to about -10.5 oC for the coldest experiment. Severe difficulties to get the same inlet conditions in the different experiments, especially regarding pressure drop, air flow rates and supply inlet temperatures, made it difficult to compare the results. However, the exhaust air side of the test rig delivered very stable humidity and temperature conditions through the test periods and were found to be almost not affected by changes in the surroundings. The hypothesis that ice formation would create an increase in the pressure drop over the exhaust air side of the heat exchanger was right according to the results. This means that the test rig was suitable to investigate if ice formation problems occurred or not. The uncertainty of the experimental results was huge, especially concerning the flow rate measurements and the temperature loss to the surroundings. The equipment used for this experimental investigation were not ideal, especially since the cooling coil tended to freeze and the air flow rate on the supply air side therefore decreased with time. The test length was restricted to the time before the cooling coil was filled with ice and blocked the air flow rate. For most experiments this happened after about 800 minutes. Nevertheless, this was enough time to see ice formation in the plastic based heat exchangers. Actually condensation of water was observed to occur after short time in these experiments (1,2 and 4). The differences in pressure drops were also starting to rice at the beginning of these experiments (see figure 4.7.). If the flow rates are equal, there is no ice formation or condensation, the heat capacity is assumed constant and there are no losses to the surroundings the lines indicating the change in temperature and humidity for the air streams will be of equal length and inclination in the temperature-humidity diagram (as shown in figure 2.2.). In the test rig it was almost impossible to get equal flow rates due to the problems in regulating this, and the difficulty in measuring it as well. The temperature efficiency was calculated based on the exhaust air side as mentioned in chapter 3.3. This lead to that in the case where the exhaust air flow rate was greater than the supply air flow rate, as in experiment 3, the line for the supply air is longer (see figure A.15). When the flow rate for the supply air was greater than the air flow for the exhaust air the line for exhaust air should have been longer than the one for supply air. However, due to experiment 2,4,5 and 6 this was not the case. The supply air lines were still longer (in experiment 4) or of the same magnitude. This may be 53

Membrane Based Heat Exchanger

.

explained from that the cold supply air experienced a greater temperature loss to the surroundings than the exhaust air. Since the surrounding temperature heated the supply air the line was longer than if it had been no loss. How great this loss was may be found from experiment 4 where the flow rates only differed by approximately 6%. The temperature loss to the surroundings based on the difference in lines length was around 10 oC for the case when the supply inlet temperature was -8.4 oC. From the experiment 5 the loss was however below 4oC.

5.4. Evaluation of the Membrane Based Heat Exchanger Prototype The plastic based heat exchanger prototypes (both the Optimax plastic sheets and the DuPont PP-material) were shown to experience water condensation and frost formation in the exhaust air channels. Water droplets were observed through the transparent acrylic top plate in the top exhaust channel already after few hours in these experiments. The frost formation appeared near the supply air inlet (see figure 4.12.) which correlates very well with the CFD analysis of Zhang (2010). Zhang’s (2010) analysis shows that the coldest area in the exhaust air channels will be near the supply air inlet. See reprinted figure below:

Figure 5.5.

The dimensionless temperature change and flow pattern in the warm air side through a quasi-counter flow heat exchanger. Reprinted overlaid figures from Zhang (2010).

The pressure drop difference in these experiments riced significantly through the test periods according to table 4.3. In the membrane based experiments 3, 5, 6 and 7 neither condensation nor ice formed. The change in pressure drop differences were not significantly (experiment 7 must be disregarded here since the test period was too short). In the last experiment a 54

Chapter 5: Discussion

small amount of condensate water was observed near the supply air outlet in the upper exhaust air channel after 8 hours. The membrane had expanded and was also crumpled in this area in this experiment. The DuPont X-membrane material tested was shown to expand in very humid conditions. See figure 4.12. At the end of the experiment ice was found as shown in picture A.41. and 4.12. Since the ice formation appeared in different areas in the plastic based prototypes contra the last membrane based experiment, the ice formation in the latter may have been caused by other means than in the previous. In the plastic based exchanger prototypes the ice formed in the area where the coldest temperature in the exchanger acted; near the supply air inlet. Conversely, the ice in the membrane based heat exchanger appeared near the supply air outlet which actually was the second warmest side of the exchanger. As shown in the picture series in figure 4.4. and 4.5. the crumpled membrane changed the air flow pattern, making the air in one part of the exchanger area stagnant. A hypothesis of why ice was formed in this part in the experiment 8 was made: The crumpling of the membrane due to high humidity levels in the exhaust air side made the membrane stuck to the upper and lower heat exchanger frame plates making the exhaust air stagnant. The supply air side cooled the stagnant exhaust air to down to (almost) the temperature of the supply air. Since the moisture transfer efficiency never can be 100% condensation and freezing occurred. Zhang’s (2010) CFD-analysis showed that the most humid area was on the heat exchangers’ lower right side (see figure below) while the ice formation in the experiment was on the lower left side. However, Zhang (2010) showed that the lower left side had a lower velocity than the right side. This can explain why the extensive crumpling and ice formation happened here in the exhaust air side. From figure 5.5. the temperature was also found to be lower in the right side than at the left, increasing the relative humidity.

55

Membrane Based Heat Exchanger

Figure 5.6.

.

Left: The dimensionless humidity change and flow pattern in the warm air side through a quasi-counter flow heat exchanger. The exhaust air inlet is at the lower rights side. Reprinted overlaid figures from Zhang (2010). Right: the ice formation and crumpling of the tested heat and moisture exchanger. (The picture is mirrored for comparison)

As neither condensate water nor ice were found in the coldest spot (near the supply air inlet) this may indicate that the expansion of the membrane was the problem that caused the ice formation. Expansion and crumpling were observed when the humidity was high and were experienced in experiment 8 with an exhaust humidity of 46.6% and supply air temperature of -9.6 oC, but not in experiment 7 with 37.3% and -10.5 oC. Experiment 8 had the highest exhaust inlet absolute humidity at 8 gW/kgA. The supply air temperature in experiment 7 was colder than in experiment 8, but the exhaust humidity was lower (6.5 gW/kgA). Therefore it seems like the temperature and humidity condition that lead to crumpling of the membranes laid somewhere in the range between these two experiments. In the picture below lines are drawn between the exhaust inlet condition and supply inlet condition for the different membrane exchanger experiments in a temperature-humidity diagram. As seen the line for the experiment 8 is much closer to the saturation line than the other experiments. This may explain the expansion of the membrane in this experiment.

56

Chapter 5: Discussion

Figure 5.7.

Lines drawn between the membrane based experiments inlet conditions in a temperature-humidity diagram.

Since the humidity inside a residential building in winter seldom gets above 40% RH pursuant to Kalamees et al. (2009). See figure 1.3. This means that the tested membrane based heat exchanger may work in even colder supply air temperatures than -10 oC in normal conditions. Figure 4.10. shows the relation between the moisture transfer efficiency on the exhaust air side and the supply inlet temperature. The relation follows the effect found in the theory from Niu and Zhang (2000) showed in figure 2.11. The colder the temperature in the supply air inlet was the higher the mean relative humidity close to the membrane. This decreased the mean moisture transfer resistance and increased the moisture transfer efficiency. Since the exhaust air flow rate was much higher than the supply air flow at the two last experiments the measured efficiencies at the exhaust air side reached 90%. This is of course a much higher value than it would have been if the air flow rates were equal. Nevertheless, the correlation shows that an even colder supply inlet temperature may lead to higher moisture transfer efficiency. It is impossible to define how much better the membrane based heat exchanger would work regarding to energy savings compared to a plastic based exchanger in a residential building use from the test performed in this work. The heat exchanger area and the flow rates were too low to compare this experiment to a “real case”. However, the experimental results shows that the freezing and condensation problems were less in the membrane based exchanger. It is need for more investigation to find out at what temperature and humidity levels the membrane based exchanger also will 57

Membrane Based Heat Exchanger

.

experience freezing due to the limitation of moisture transfer effectiveness and not because of expansion of the membrane due to humidity levels. The membrane tested had drawbacks as it was very elastic which made it difficult to get the membranes stretched out when building the heat exchanger prototype. The pressure prop behaviour cause by the elasticity discussed in chapter 5.2 may cause problems if the flow rates are unbalanced. The elastic membranes would probably create an even bigger unbalance and the energy needed for pumping the air through the heat exchanger would increase. The optimal membrane should therefore not expand when soaked in water and should preferably not be elastic due to the problems if the flow rates get imbalanced and the difficulty in building the exchanger.

58

Chapter 6: Conclusion and Further Work

Chapter 6

Conclusion and Further Work The mathematical method derived to predict the moisture transfer effectiveness in a membrane based heat exchanger was shown to fit very well with the experimental results. Utilising available permeability test results to find an average moisture transfer resistance were shown to be an appropriate method to find the overall moisture transfer coefficient. The direct computational method was simple to implement in a calculation tool since it did not require any iteration processes or additional testing of the membrane. The calculation tool HXcalc made in Microsoft Excel may be used for further work to investigate the appropriateness for different membranes in a membrane based heat exchanger. The experimental tests showed that the heat exchanger prototypes utilizing plastic sheets, as the heat transfer material, experienced condensation and ice formation in the exhaust air channels near the supply air inlet side of the exchanger. In the experiments utilising the hydrophilic membrane X from DuPont as the heat and moisture transfer material neither condensation nor ice were found near the supply air inlet. However in the experiment with the highest exhaust inlet humidity (46.6% RH) the membrane had expanded and was crumpled near the supply air outlet. Condensate water and ice were found in the exhaust air channels near the supply air outlet in this experiment. The hydrophilic membrane X from DuPont was therefore found to be superior to the two plastic materials regarding water condensation and frost formation in the heat exchanger prototype when the exhaust air relative humidity was below 37% and the temperature above -10.5 oC. The pressure drops over the heat exchanger were found to be strongly influenced by the membrane material’s elasticity and were not proportional to the flow rate as expected. The elasticity and the membrane’s tendency to expand at high humidity made the tested membrane unsuitable for use in a membrane based heat exchanger. Methods to support the membrane to decrease the elasticity should be investigated. Lamination of the membrane to a supporting fabric may be a solution that should be tested. Other types of membranes should also be tested. The experimental investigation was restricted to a supply air temperature at about -10 o C at the lowest. The membrane based heat exchangers performance at even lower temperatures should be investigated to see if the membrane based heat exchanger could work in extreme winter conditions without defrosting systems (frost guards etc). A plastic based heat exchanger should be tested more carefully to find the lowest exhaust air humidity and warmest supply air temperature that would lead to 59

Membrane Based Heat Exchanger

.

condensation and ice formation respectively. Then the two types of heat exchangers may be compared to find the possible energy savings from replacing plastic sheets with membranes. If the built test rig is to be used for further testing the axial fans should be replaced with centrifugal fans to make sure that the flow rates are stable even if the pressure drop over the exchanger changes due to condensation and ice formation. The heat exchanger prototypes should also be rebuilt with an increased number of heat exchanger plates to be able to increase the air flow rates still remaining high efficiency. Higher air flow rates will decrease the temperature loss to the surroundings and there will be possible to test the heat exchanger at colder supply inlet temperatures. The membrane should also be tested for durability and pollution transfer to decide if the technology is suitable for use in residential buildings with several living units.

60

References

References Aarnes, S. 2011. Energy-Efficient Ventilation Solutions for ZEB. Project Thesis, NTNU, Trondheim. Alves, V. D., & Coelhoso, I. M. 2004. Effect of membrane characteristics on mass and heat transfer in the osmotic evaporation process. Journal of Membrane Science, 228(2): 159-167. ASTM. 2011. E96/96M-10 Standard Test Methods for Water Vapor Transmission of Materials. West Conshohocken, PA: ASTM International. Chen, L.-C., Yu, T. L., et al. 2008. Nafion/PTFE and zirconium phosphate modified Nafion/PTFE composite membranes for direct methanol fuel cells. Journal of Membrane Science, 307(1): 10-20. Crank, J. 1975. The mathematics of diffusion. Oxford: Clarendon Press. Drivsholm, C., Olsen, H., Larsen, C. G., Jensen, J. S., Nielsen, T. R., Kragh, J., & Svendsen, S. 2005. Udvikling af energiøkonomisk ventilationsløsning med varmegenvinding til boliger: BYG-DTU. Favre, E., Clement, R., Nguyen, Q. T., Schaetzel, P., & Neel, J. 1993. Sorption of organic solvents into dense silicone membranes. Part 2.-Development of a new approach based on a clustering hypothesis for associated solvents. Journal of the Chemical Society, Faraday Transactions, 89(24). Gibson, P. W. 2000. Effect of temperature on water vapor transport through polymer membrane laminates. Polymer Testing, 19(6): 673-691. Huang, J., Cranford, R. J., et al. 2004. Sorption and transport behavior of water vapor in dense and asymmetric polyimide membranes. Journal of Membrane Science, 241(2): 187-196.effectiveness Huizing, R. N. 2010. Coated Membranes for Enthalpy Exchange and Other Applications. US Patent and Trademark Office. US Patent PCT/CA2010/000735 Incropera, F. P., & DeWitt, D. P. 2007. Fundamentals of heat and mass transfer: John Wiley. ISO. 2001. NS-EN ISO 12572: Hygrothermal performance of building materials and products- Determination of water vapour transmission properties. Oslo, Norway: Standard Norge.

61

Membrane Based Heat Exchanger

.

JSA. 1999. JIS L 1099: Testing Methods for water Vapour Permeability of Clothes. Tokyo, Japan: Japanese Standard Association. Kadylak, D., Cave, P., et al. 2009. Effectiveness correlations for heat and mass transfer in membrane humidifiers. International Journal of Heat and Mass Transfer, 52(5-6): 1504-1509. Kalamees, T., Korpi, M., Vinha, J., & Kurnitski, J. 2009. The effects of ventilation systems and building fabric on the stability of indoor temperature and humidity in Finnish detached houses. Building and Environment, 44(8): 1643-1650. Kays, W. M., & London, A. L. 1964. Compact heat exchangers. New York: McGrawHill. Marais, S., Métayer, M., et al. 2000. Permeametric and microgravimetric studies of sorption and diffusion of water vapor in an unsaturated polyester. Polymer, 41(7): 2667-2676. McCullough, E. A., Myoungsook, K., & Huensup, S. 2003. A comparison of standard methods for measuring water vapour permeability of fabrics. Measurement Science and Technology, 14(8): 1402. Modesti, M., Dall’Acqua, C., et al. 2004. Mathematical model and experimental validation of water cluster influence upon vapour permeation through hydrophilic dense membrane. Journal of Membrane Science, 229(1–2): 211223. Mondal, S., Hu, J. L., et al. 2006. Free volume and water vapor permeability of dense segmented polyurethane membrane. Journal of Membrane Science, 280(1–2): 427-432. Mukhopadhyay, A., & Midha, V. K. 2008. A Review on Designing the Waterproof Breathable Fabrics Part I: Fundamental Principles and Designing Aspects of Breathable Fabrics. Journal of Industrial Textiles, 37(3): 225-262. Nasif, M. S., Morrison, G. L., & Behnia, M. 2005. Heat and Mass Transfer in Air to Air Enthalpy Heat Exchangers. 6th World Conference on Experimental Heat Transfer, Fluid Mechanics, and Thermodynamics. April 17-21, 2005, Matsushima, Miyagi, Japan Niu, J. L., & Zhang, L. Z. 2000. Membrane-based Enthalpy Exchanger: material considerations and clarification of moisture resistance. Journal of Membrane Science, 189(2): 179-191. Novakovic, V., Hansen, S. O., Thue, J. V., & Gjerstad, F. O. 2007. ENØK i bygninger: effektiv energibruk. Oslo: Gyldendal undervisning.

62

References

Ouazia, B. K., Swinton, M. C., Julien, M., & Manning, M. 2006. Assessment of the enthalpy performance of houses using energy recovery technology. Paper presented at the 2006 Winter Meeting of the American Society of Heating, Refrigerating and Air-Conditioning Engineers, ASHRAE, January 21, 2006 January 25, 2006, Chicago, IL, United states. Peron, J., Mani, A., et al. 2010. Properties of Nafion® NR-211 membranes for PEMFCs. Journal of Membrane Science, 356(1–2): 44-51. Vali, A., Simonson, C. J., Besant, R. W., & Mahmood, G. 2009. Numerical model and effectiveness correlations for a run-around heat recovery system with combined counter and cross flow exchangers. International Journal of Heat and Mass Transfer, 52(25-26): 5827-5840. Wijmans, J. G., & Baker, R. W. 1995. The solution-diffusion model: a review. Journal of Membrane Science, 107(1–2): 1-21. Yeh, H. M., & Chang, Y. H. 2005. Mass transfer for dialysis through parallel-flow double-pass rectangular membrane modules. Journal of Membrane Science, 260(1-2): 1-9. Yoshino, M., & Hashimoto, Y. 1973. The Lossnay penetration-type total heat exchanger. Mitsubishi Electric Engineer(38): 24-27. Yun, R., Kim, Y., & Min, M.-k. 2002. Modeling of frost growth and frost properties with airflow over a flat plate. International Journal of Refrigeration, 25(3): 362-371. Zhang, L. Z., & Niu, J. L. 2001. Energy requirements for conditioning fresh air and the long-term savings with a membrane-based energy recovery ventilator in Hong Kong. Energy, 26(2): 119-135. Zhang, L. Z., & Niu, J. L. 2002. Effectiveness Correlations for Heat and Moisture Transfer Processes in an Enthalpy Exchanger With Membrane Cores. Journal of Heat Transfer, 124(5): 922-929. Zhang, L.-Z. 2010. Heat and mass transfer in a quasi-counter flow membrane-based total heat exchanger. International Journal of Heat and Mass Transfer, 53(23-24): 5478-5486. Zhang, L.-Z. 2012. Progress on heat and moisture recovery with membranes: From fundamentals to engineering applications. Energy Conversion and Management(0). http://dx.doi.org/10.1016/j.enconman.2011.11.033

63

Appendix A. A.

B.

C.

A

A.1.

Initial Testing of Cooling Coils

A-3

A.2.

Development of Stable Exhaust Air Conditions

A-5

A.3.

Program code

A-7

A.4.

Results

A-13

A.4.1 Experiment 1: Three layer Optimax plastic sheets

A-14

A.4.2 Experiment 2: Five layer Optimax plastic sheets

A-17

A.4.3 Experiment 3: DuPont X-membrane

A-20

A.4.4 Experiment 4: DuPont PP

A-23

A.4.5 Experiment 5: DuPont X

A-26

A.4.6 Experiment 6: DuPont X

A-29

A.4.7 Experiment 7: DuPont X

A-32

A.4.8 Experiment 8: DuPont X

A-35

A.5.

Heat Exchanger Prototype: Mechanical Drawing

A-38

A.6.

HSE Report

A-39

.

Appendix.

A.1. Initial Testing of Cooling Coils Testing of the cooling system already installed in the VVS-lab at EPT was carried out initially to see if the system could be used for the experiment. It was preferable to use the installed system to save time and money. The cooling coil was first tested with outdoor air and quite high air flow rates given by the natural drag through the ducting system. Extensive frost formation on the cooling coils was experienced and the flow rate was decreased after about 6 hours. After 12 hours the cooling coil outlet was completely blocked with frost. Second another approach was tried with smaller air flow rates through the cooling device. A small fan was placed in front of the inlet duct. Warm indoor air from the lab was used to test the “worst case scenario” of frost formation inside the coil. First the cooling coil was tested for 24 hours. The flow rate was measured with the Velocicalc 9555-P instrument from TSI by measuring the velocity in three different points in the outlet channel and average the value. The flow rate was adjusted by changing the outlet area of the outlet duct using airtight “duct tape”. The Velocicalc instrument was used to log the pressure drop over the cooling coil and the temperature and humidity after the cooling coil. After 24 hours no increasing pressure drop was found. The outlet humidity was rising through the test period. Since the inlet humidity and temperature was unknown the reason for increasing outlet humidity was unknown. Another test period of 24 hours was then carried out. The cooling coil and fan was not turned off in the break between the test periods. The two tests may therefore be seen in connection. In the second test period the inlet air humidity and temperature was measured using a TinyTag instrument from Gemini.

Figure A.1.

Internal testing of cooling coil. Temperature, relative humidity and pressure were tested for a 24 hour period first. For the second 24 hour period the temperature and humidity in the surrounding room. The outlet temperature was pretty stable at about -12oC and the humidity was between 45-55% through the tests.

After approximately 38 hours the pressure started to increase. At the same time the outlet temperature started to increase and the outlet humidity started to decrease. The outlet temperature and humidity did not correlate with the fluctuations in the inlet temperature and humidity.

A-3

Membrane Based Heat Exchanger

.

The test showed that the humidity increased through the whole test period until the pressure drop started to increase. Nevertheless since the outlet temperature was quite low the increase in relative humidity will correspond to a very low change in absolute humidity. The inertial testing of the cooling coil seemed therefore promising for tests up to 38 hours. The cooling coil had been turned on for approximately two hours before the measurements begun. Why the humidity didn’t draw near 100% may indicate that the air in fact was colder somewhere inside the cooling coli than at the outlet were the measurement was taken. The small air flow rates may have caused almost stagnant air in the outlet duct witch then may have had a great temperature exchange with the warm surroundings. Insulation of the outlet side of the cooling device was then needed.

A-4

Appendix.

A.2. Development of Stable Exhaust Air Conditions Different attempts were tried out to make warm and humid exhaust air that were stable over a long time. In this appendix is two attempts that were rejected shortly described. The final method is described in chapter 3. A simple system without too many advanced control mechanisms was preferable. A polystyrene box was made to have the heat and humidifier equipment inside. This made it easy to change refill water in the humidifier and the box acted as a small “climate-chamber”. Try-out 1: Electrical humidifier and fan heater. An “ultrasonic” humidifier was put inside the box together with a 2000 W electrical fan heater with an inbuilt thermostat. Pretesting showed a great fluctuation in the humidity level. The temperature level did also fluctuate to some extent. The temperature and humidity were masured by the Velocicalc 9555-P from Vaisala.

Figure A.2.

Overview of a first attempt to build attest rig. Drawing made in Google Sketch Up. The exhaust air was heated and humidified in a polystyrene box placed under a table to make the rig as compact as possible.

A-5

Membrane Based Heat Exchanger

Figure A.3.

.

Internal testing of exhaust air side. The temperature seemed quite stable while the humidity level changed greatly through the test period.

Try-out 2: fan heater and water bracket. A second approach was tried with an open bracket of water put inside the polystyrene box together with the fan heater. There were still observed fluctuations in the humidity level. This sudden drops in the humidity level corresponded to the on and of switching of the fan heater that probably changed the flow conditions in the polystyrene box a great deal.

Figure A.4.

Internal testing of exhaust air side. The temperature seemed quite stable while the humidity level still fluctuating.

A-6

Appendix.

A.3. Program code Sub inputvar() Call clearsheet2 Min = Worksheets(1).Range("c5") Max = Worksheets(1).Range("c6") Begin = Min If Worksheets(2).Range("d11") = 1 Then Worksheets(2).Range("b43").Value Worksheets(2).Range("b44").Value Worksheets(2).Range("b45").Value Worksheets(2).Range("b46").Value Worksheets(2).Range("b47").Value Worksheets(2).Range("b49").Value Ant = (Max - Min) / 2 Teller = 2 Varcell = 48

= = = = = =

ElseIf Worksheets(2).Range("d11") = 2 Then Worksheets(2).Range("b43").Value = Worksheets(2).Range("b44").Value = Worksheets(2).Range("b46").Value = Worksheets(2).Range("b47").Value = Worksheets(2).Range("b48").Value = Ant = Worksheets(1).Range("c7") Teller = (Max - Min) / (Ant - 1) Varcell = 45

Worksheets(1).Range("c26") Worksheets(1).Range("c27") Worksheets(1).Range("c28") Worksheets(1).Range("c29") Worksheets(1).Range("c30") Worksheets(1).Range("c31")

Worksheets(1).Range("c26") Worksheets(1).Range("c27") Worksheets(1).Range("c29") Worksheets(1).Range("c30") Worksheets(1).Range("c21")

ElseIf Worksheets(2).Range("d11") = 3 Then Worksheets(2).Range("b43").Value = Worksheets(1).Range("c26") Worksheets(2).Range("b44").Value = Worksheets(1).Range("c27") Worksheets(2).Range("b45").Value = Worksheets(1).Range("c28") Worksheets(2).Range("b47").Value = Worksheets(1).Range("c30") Worksheets(2).Range("b48").Value = Worksheets(1).Range("c21") Worksheets(2).Range("b49").Value = Worksheets(1).Range("c31") Ant = Worksheets(1).Range("c7") Teller = (Max - Min) / (Ant - 1) Varcell = 46 ElseIf Worksheets(2).Range("d11") = 4 Then Worksheets(2).Range("b44").Value = Worksheets(1).Range("c27") Worksheets(2).Range("b45").Value = Worksheets(1).Range("c28") Worksheets(2).Range("b46").Value = Worksheets(1).Range("c29") Worksheets(2).Range("b47").Value = Worksheets(1).Range("c30") Worksheets(2).Range("b48").Value = Worksheets(1).Range("c21") Worksheets(2).Range("b49").Value = Worksheets(1).Range("c31") Ant = Worksheets(1).Range("c7") Teller = (Max - Min) / (Ant - 1) Varcell = 43 ElseIf Worksheets(2).Range("d11") = 5 Then Worksheets(2).Range("b43").Value = Worksheets(1).Range("c26") Worksheets(2).Range("b45").Value = Worksheets(1).Range("c28") Worksheets(2).Range("b46").Value = Worksheets(1).Range("c29") Worksheets(2).Range("b47").Value = Worksheets(1).Range("c30") Worksheets(2).Range("b48").Value = Worksheets(1).Range("c21") Worksheets(2).Range("b49").Value = Worksheets(1).Range("c31") Ant = Worksheets(1).Range("c7") Teller = (Max - Min) / (Ant - 1) Varcell = 44 ElseIf Worksheets(2).Range("d11") = 6 Then Worksheets(2).Range("b43").Value = Worksheets(1).Range("c26") Worksheets(2).Range("b44").Value = Worksheets(1).Range("c27") Worksheets(2).Range("b45").Value = Worksheets(1).Range("c28") Worksheets(2).Range("b46").Value = Worksheets(1).Range("c29") Worksheets(2).Range("b48").Value = Worksheets(1).Range("c21") Worksheets(2).Range("b49").Value = Worksheets(1).Range("c31") Ant = Worksheets(1).Range("c7") Teller = (Max - Min) / (Ant - 1)

A-7

Membrane Based Heat Exchanger

.

Varcell = 47 End If Worksheets(2).Cells(Varcell, 2).Value = Begin For K = 1 To Ant Call Fmini Fmin = Worksheets(2).Range("h46") Fe = Worksheets(2).Range("i46") Fs = Worksheets(2).Range("j46") If Worksheets(2).Range("d32") = 1 Then Call geomcross If Worksheets(2).Range("d17") = 1 Then Call membranecross End If ElseIf Worksheets(2).Range("d32") = 2 Then Call geomcounter If Worksheets(2).Range("d17") = 1 Then Call membranecounter End If ElseIf Worksheets(2).Range("d32") = 3 Then Call geomcross Call geomcounter Atot = Worksheets(2).Range("e52") + Worksheets(2).Range("e53") nT = (Worksheets(2).Range("e53") / Atot) * Worksheets(2).Range("k53") + (Worksheets(2).Range("e52") / Atot) * Worksheets(2).Range("k52") Worksheets(2).Range("k54").Value = nT DP = Worksheets(2).Range("i52") + Worksheets(2).Range("i53") Worksheets(2).Range("i54").Value = DP Worksheets(2).Range("e54").Value = Atot If Worksheets(2).Range("d17") = 1 Then Call membranecross Call membranecounter nM = (Worksheets(2).Range("e53") / Atot) * Worksheets(2).Range("m53") + (Worksheets(2).Range("e52") / Atot) * Worksheets(2).Range("m52") Worksheets(2).Range("m54").Value = nM End If End If 'Outoput: Tein = Worksheets(2).Range("b43") Tsin = Worksheets(2).Range("b46") øein = Worksheets(2).Range("b44") øsin = Worksheets(2).Range("b47") wein = øein * 10 ^ 7 / (6.19 * Exp(5427 / (273.15 + Tein))) wsin = øsin * 10 ^ 7 / (6.19 * Exp(5427 / (273.15 + Tsin))) Worksheets(2).Range("c44").Value = wein Worksheets(2).Range("c47").Value = wsin Worksheets(2).Range("b60").Value = xax Worksheets(2).Cells(60 + K, 2).Value = Worksheets(2).Cells(Varcell, 2) Hxgeom = Worksheets(2).Range("d32") Worksheets(2).Cells(60 + K, 3).Value = Worksheets(2).Cells(51 + Hxgeom, 11) Worksheets(2).Cells(60 + K, 6).Value = Worksheets(2).Cells(60 + K, 3) * (Tein - Tsin) + Tsin Worksheets(2).Cells(60 + K, 4).Value = Worksheets(2).Cells(51 + Hxgeom, 13) Worksheets(2).Cells(60 + K, 7).Value = -Worksheets(2).Cells(60 + K, 3) * (Tein - Tsin) + Tein Worksheets(2).Cells(60 + K, 8).Value = Worksheets(2).Cells(60 + K, 4) * (wein - wsin) + wsin Worksheets(2).Cells(60 + K, 9).Value = -Worksheets(2).Cells(60 + K, 4) * (wein - wsin) + wein Worksheets(2).Cells(60 + K, 10).Value = Tsin Worksheets(2).Cells(60 + K, 11).Value = Tein Worksheets(2).Cells(60 + K, 12).Value = wsin Worksheets(2).Cells(60 + K, 13).Value = wein Tfrost = 5427 / (Log(10 ^ 7 / (6.19 * Worksheets(2).Cells(60 + K, 9)))) - 273 Ta = Tein - (Tein - Tfrost) / (Worksheets(2).Cells(60 + K, 3)) Worksheets(2).Cells(60 + K, 14).Value = Ta Call recoveryefficiency Worksheets(2).Cells(60 + K, 15).Value = Worksheets(2).Cells(51 + Hxgeom, 16) Worksheets(2).Cells(60 + K, 5).Value = Worksheets(2).Cells(51 + Hxgeom, 15) Worksheets(2).Cells(60 + K, 16).Value = Worksheets(2).Cells(60 + K, 5) * Worksheets(2).Cells(60 + K, 15) Begin = Begin + Teller Worksheets(2).Cells(Varcell, 2).Value = Begin Worksheets(2).Cells(60 + K, 20).Value = Worksheets(2).Cells(51 + Hxgeom, 9)

A-8

Appendix. Next Worksheets(2).Range("b58").Value = Ant End Sub

Sub geomcross() h = Worksheets(1).Range("e20") Channelheight = h / (Worksheets(2).Range("b48") + 1) W = Worksheets(1).Range("f20") Dh = 2 * Channelheight * W / (Channelheight + W) L = W A = Channelheight / W Atot = L * W * (Worksheets(2).Range("b48")) Visc = Worksheets(1).Range("c37") rho = Worksheets(1).Range("c36") kond = Worksheets(1).Range("c35") cp = Worksheets(1).Range("c38") Across = Channelheight * W Fmin = Worksheets(2).Range("h46") Fe = Worksheets(2).Range("i46") Fs = Worksheets(2).Range("j46") R = Worksheets(2).Range("k46") Vel = Fmin / ((h / 2) * W) Re = Vel * Dh / Visc f = (-50.416 * A ^ 3 + 132.75 * A ^ 2 - 121.22 * A + 95.705) / Re K = 4 * f * L / Dh DP = 0.5 * rho * Vel ^ 2 * (K + 0.8775) Nu = -7.4818 * (A ^ 3) + 18.535 * (A ^ 2) - 15.663 * A + 8.235 Worksheets(2).Range("g42").Value = Nu hc = Nu * kond / Dh U = (2 / hc) ^ -1 NTU = U * Atot / (cp * Fmin * rho) nT = (1 - Exp((1 / R) * (NTU ^ 0.22) * (Exp(-R * (NTU ^ 0.78)) - 1))) / (Fe / Fmin) Worksheets(2).Range("b52").Value Worksheets(2).Range("c52").Value Worksheets(2).Range("d52").Value Worksheets(2).Range("e52").Value Worksheets(2).Range("f52").Value Worksheets(2).Range("g52").Value Worksheets(2).Range("h52").Value Worksheets(2).Range("i52").Value Worksheets(2).Range("j52").Value Worksheets(2).Range("k52").Value End Sub

= = = = = = = = = =

Dh NTU L Atot A Vel Re DP hc nT

Sub geomcounter() h = Worksheets(1).Range("e20") Channelheight = h / (Worksheets(2).Range("b48") + 1) W = Worksheets(1).Range("h20") Dh = 2 * Channelheight * W / (Channelheight + W) L = Worksheets(1).Range("g20") A = Channelheight / W Atot = L * W * (Worksheets(2).Range("b48")) Visc = Worksheets(1).Range("c37") rho = Worksheets(1).Range("c36") kond = Worksheets(1).Range("c35") cp = Worksheets(1).Range("c38") Fmin = Worksheets(2).Range("h46") Fe = Worksheets(2).Range("i46") Fs = Worksheets(2).Range("j46") R = Worksheets(2).Range("k46")Worksheets(2).Range("f42").Value = R Across = Channelheight * W Vel = Fmin / ((h / 2) * W) Re = Vel * Dh / Visc f = (-50.416 * A ^ 3 + 132.75 * A ^ 2 - 121.22 * A + 95.705) / Re K = 4 * f * L / Dh DP = 0.5 * rho * Vel ^ 2 * (K + 0.8775) Nu = -7.4818 * A ^ 3 + 18.535 * A ^ 2 - 15.663 * A + 8.235 Worksheets(2).Range("g43").Value = Nu hc = Nu * kond / Dh U = (2 / hc) ^ -1 NTU = U * Atot / (cp * Fmin * rho) If R < 1 Then

A-9

Membrane Based Heat Exchanger

.

nT = ((1 - Exp(-NTU * (1 - R))) / (1 - R * Exp(-NTU * (1 - R)))) / (Fe / Fmin) ElseIf R = 1 Then nT = (NTU / (1 + NTU)) / (Fe / Fmin) End If Worksheets(2).Range("b53").Value Worksheets(2).Range("c53").Value Worksheets(2).Range("d53").Value Worksheets(2).Range("e53").Value Worksheets(2).Range("f53").Value Worksheets(2).Range("g53").Value Worksheets(2).Range("h53").Value Worksheets(2).Range("i53").Value Worksheets(2).Range("j53").Value Worksheets(2).Range("k53").Value End Sub

= = = = = = = = = =

Dh NTU L Atot A Vel Re DP hc nT

Sub membranecross() 'equation 2.8. hm = Worksheets(2).Range("j52") / (Worksheets(1).Range("c38") * (Worksheets(1).Range("c16")) ^ (2 / 3)) Rconv = 2 / (hm) Fmin = Worksheets(2).Range("h46") Fe = Worksheets(2).Range("i46") Fs = Worksheets(2).Range("j46") R = Worksheets(2).Range("k46") 'reference resistances R1 = (Worksheets(1).Range("c13")) r2 = (Worksheets(1).Range("c15")) Ø1 = Worksheets(1).Range("g12") T1 = Worksheets(1).Range("f12") Ø2 = Worksheets(1).Range("g14") K = (r2 / R1 - 1) / (Ø1 / Ø2 - 1) øein = Worksheets(2).Range("b44") Tein = Worksheets(2).Range("b43") øsin = Worksheets(2).Range("b47") Tsin = Worksheets(2).Range("b46") wmax = øein * 10 ^ 7 / (6.19 * Exp(5427 / (273.15 + Tein))) wmin = øsin * 10 ^ 7 / (6.19 * Exp(5427 / (273.15 + Tsin))) wref = Ø1 * 10 ^ 7 / (6.19 * Exp(5427 / (273.15 + T1))) wmean = (wmin + wmax) / 2 Tmean = (Tein + Tsin) / 2 ømean = wmean / (10 ^ 7 / (6.19 * Exp(5427 / (273.15 + Tmean)))) 'mean resistance: Rmem = R1 * (1 - K + K * (Ø1 / (ømean))) Um = 1 / (Rconv + Rmem) NTUm = Um * (Worksheets(2).Range("e52")) / (Fmin * Worksheets(1).Range("c36")) Worksheets(2).Range("n52").Value = NTUm 'moisture transfer efficiency from 2.4 nM = (1 - Exp(NTUm ^ 0.22 * (Exp(-R * NTUm ^ 0.78) - 1) / R)) / (Fe / Fmin) Worksheets(2).Range("l52").Value = Rmem Worksheets(2).Range("m52").Value = nM Worksheets(2).Range("f45").Value = hm End Sub

Sub membranecounter() hm = Worksheets(2).Range("j53") / (Worksheets(1).Range("c38") * (Worksheets(1).Range("c16")) ^ (2 / 3)) Rconv = 2 / (hm) Fmin = Worksheets(2).Range("h46") Fe = Worksheets(2).Range("i46") Fs = Worksheets(2).Range("j46") R = Worksheets(2).Range("k46") R1 = (Worksheets(1).Range("c13")) r2 = (Worksheets(1).Range("c15")) Ø1 = Worksheets(1).Range("g12") Ø2 = Worksheets(1).Range("g14") T1 = Worksheets(1).Range("f12") K = (r2 / R1 - 1) / (Ø1 / Ø2 - 1) øein = Worksheets(2).Range("b44") Tein = Worksheets(2).Range("b43") øsin = Worksheets(2).Range("b47")

A-10

Appendix. Tsin = Worksheets(2).Range("b46") wmax = øein * 10 ^ 7 / (6.19 * Exp(5427 / (273.15 + Tein))) wmin = øsin * 10 ^ 7 / (6.19 * Exp(5427 / (273.15 + Tsin))) wref = Ø1 * 10 ^ 7 / (6.19 * Exp(5427 / (273.15 + T1))) wmean = (wmin + wmax) / 2 Tmean = (Tein + Tsin) / 2 ømean = wmean / (10 ^ 7 / (6.19 * Exp(5427 / (273.15 + Tmean)))) Rmem = R1 * (1 - K + K * (Ø1 / (ømean))) Um = 1 / (Rconv + Rmem) NTUm = (Um * Worksheets(2).Range("e53")) / (Fmin * Worksheets(1).Range("c36")) 'moiure transfer efficiency from eq 2.5 If R < 1 Then nM = ((1 - Exp(-NTUm * (1 - R))) / (1 - R * Exp(-NTUm * (1 - R)))) / (Fe / Fmin) ElseIf R = 1 Then nM = (NTUm / (1 + NTUm)) / (Fe / Fmin) End If Worksheets(2).Range("l53").Value = Rmem Worksheets(2).Range("m53").Value = nM Worksheets(2).Range("n53").Value = NTUm End Sub

Sub recoveryefficiency() Tsupply = Worksheets(1).Range("c43") Tein = Worksheets(2).Range("b43") wein = Worksheets(2).Range("c44") wsin = Worksheets(2).Range("c47") V = Worksheets(2).Range("b45") * 3600 Cpair = Worksheets(1).Range("c38") Rhoair = Worksheets(1).Range("c36") j = Worksheets(2).Range("d32") nM = Worksheets(2).Cells(51 DP = Worksheets(2).Cells(51 'get nT nT = Worksheets(2).Cells(51 sumwithout = 0 DPsum = 0 summe = 0 'Set Ta and Tb If nM > 0 Then weut = -nM * (wein Else weut = wein End If

+ j, 13) + j, 9) + j, 11)

wsin) + wein

Tfrost = -8 Ta = Tein - (Tein - Tfrost) / nT If Ta < -273 Then Ta = -273 End If Tb = 'sums place For K

nT * (Tein - Ta) + Ta hourly = Worksheets(2).Range("k16") = 2 To 8761 Tsupply = Worksheets(1).Range("c43") Tsin = Worksheets(2).Cells(K, 16 + place) Tc = nT * (Tein - Tsin) + Tsin 'Set by-pass in summer B = 0 If Tsin > Tsupply Then B = 1 End If 'find heat needed without recovery sumwithout = sumwithout + (1 - B) * Rhoair * V * Cpair * (Tsupply - Tsin) 'find energy based on pressue drop both air streams. No pressure drop summer DPsum = DPsum + 2 * DP * V * (1 - B) 'Sets Dtfrost If (Ta - Tsin) > 0 Then DTfrost = Ta - Tsin Else DTfrost = 0 End If 'Sets Dtsupply

A-11

Membrane Based Heat Exchanger

.

If Tsupply > Tc Then Dta = Tc Else Dta = Tsupply End If If Tb > Dta Then Dtb = Tb Else Dtb = Dta End If If (Tsupply - Dtb) > 0 Then Dtsupply = Tsupply - Dtb Else Dtsupply = 0 End If 'sums Dt values summe = summe + DTfrost + Dtsupply Next 'energy for heating rest after rec and frost protection Qrec = Cpair * Rhoair * summe * V 'Finds annual heat recovery efficiency nE = 1 - (Qrec + DPsum) / (sumwithout) Worksheets(2).Cells(51 + j, 15) = nE End Sub

Sub Fmini() 'Flow min If Worksheets(2).Range("d11") = 2 Then Fmin = Worksheets(2).Range("b45") Fe = Fmin Fs = Fmin R = 1 Else If Worksheets(2).Range("b45") >= Worksheets(2).Range("b49") Then Fmin = Worksheets(2).Range("b49") Fe = Fmin Fs = Worksheets(2).Range("b45") R = Fmin / Worksheets(2).Range("b45") Else Fmin = Worksheets(2).Range("b45") Fe = Worksheets(2).Range("b49") Fs = Worksheets(2).Range("b45") R = Fmin / Worksheets(2).Range("b49") End If End If Worksheets(2).Range("h46").Value Worksheets(2).Range("i46").Value Worksheets(2).Range("j46").Value Worksheets(2).Range("k46").Value End Sub

= = = =

Fmin Fe Fs R

A-12

Appendix.

A.4. Results In this chapter all results from all eight experiments are presented individually. A short description about the performed experiment comes first. Second a table showing the standard deviation, random uncertainty, mean value, systematic uncertainty and the total uncertainty for all measured and calculated results. The mean input and output temperatures and humidity levels are plotted in a humidity-temperature diagram showing the changes through the heat exchangers in each air stream. All relative humidity, temperatures, pressure drop developments and efficiency(ies) with time are displayed. For some experiments the surrounding temperature and humidity level were measures with the TSI- instrument, in other experiments the supply outlet velocity were measured with the same instruments. In these cases the results are plotted against time. For experiment 1, 4 and 8 pictures showing the heat exchanger with ice and condensate water are displayed as well. The colours used in the plots are as shown here:

Figure A.5.

Overview of the Vaisala humidity and temperature probes placement and the air streams. The colour indicates which measurement points that are displayed in the graphs later in this chapter.

A-13

Membrane Based Heat Exchanger

.

A.4.1 Experiment 1: Three layer Optimax plastic sheets The first experiment was done with a three layer plastic sheet exchanger. The plastic sheets used were transparent overhead plastic sheets from Optifilm. Only the flow rate on the supply air side was measured. The pressure drop on the supply air side was quite stable while the exhaust air side experienced a linear increase in the pressure drop. After 900 min an increase in the surrounding temperature and decrease in surrounding humidity made the inlet and outlet temperatures increase, while the humidity levels still were quite stable. At the end of the experiment dense ice had built up in the upper exhaust air channel near the supply air inlet. In the middle of the exchanger area condensed water had built up, making the upper and middle plastic sheets sticking together. In the lower channel no ice was found, however water droplets were present.

Table A.1.

Data from experiment 1. Standard deviation, random uncertainty, mean value, systematic uncertainty and total uncertainty are shown.

HMP 1 HMP 1 RH Temp o

C

% s Ur US UT

o

C

%

3.0150 0.6133 0.0112 0.0023 43.6163 23.8101 2.0000 0.1000 2.0000

HMP 2 HMP 2 RH Temp

0.1000

o

C

%

0.6344 0.0024 27.4885 2.0000

0.4696 0.0017 -5.2667 0.1000

2.0000

0.1000

kgW/kgM

HMP 3 HMP 3 RH Temp

HMP 4 RH

1.4000

kgW/kgM kgW/kgM

0.2000

Pressure Pressure Exthaust Supply

o

Pa

C

%

0.2597 0.6735 0.0010 0.0025 9.4386 11.5200 1.4000 0.2000

HMP 4 Temp 0.6738 0.0025 15.9736 0.1000

0.0621 0.0002 2.5518 0.1900

0.2065 0.0008 2.4006 0.1900

2.0001

0.1000

0.1900

0.1900

Exhaust flow Supply flow 3 m3/h kgW/kgM m /h

s Ur US UT

0.2693

0.0132

0.0079

0.0007

0.0008

0.0081 ??

1.5789

0.0044

0.0179

0.0004

0.0001

0.0001

0.0002 ??

0.1104

A-14

Pa

4.2812 0.0159 71.1096 2.0000

Appendix.

Figure A.6.

Exhaust (red) and supply (blue) air flows through the heat exchanger (based on inlet and outlet values) in a Humidity-temperature diagram. The green lines are relative humidity lines starting at 10% to the left and 100% (saturation line) at the right.

Figure A.7.

Relative humidity (left) and temperature (right) measurements through the test period in all four measurement points.

Figure A.8.

Pressure drop for both air flows. Linear trend line for the exhaust air stream is displayed.

A-15

Membrane Based Heat Exchanger

.

Figure A.9.

Left: surrounding temperature (red line) and Relative humidity (green line) over the test period. Right: Temperature efficiency.

Figure A.10.

Pictures after experiment. Left: dense ice was found in the upper right part (near the supply air inlet). In the middle a pool of water made the two upper plastic sheets stick together. Upper right: side view of the dry supply air channels. Lower right: picture inside the ice and water filled exhaust air side.

A-16

Appendix.

A.4.2 Experiment 2: Five layer Optimax plastic sheets In the second experiment the heat transfer sheets from experiment 1 were increased to five layers. This were done to increase the temperature efficiency and to see if this increased the ice formation. The flow rate at the exhaust air side were only measured once so no uncertainty were able to be found for that single measurement. Now a more significant ice layer built up in two out of three air channels in the exhaust air side. (Unfortunately no pictures were taken.)

Table A.2.

Data from experiment 2. Standard deviation, random uncertainty, mean value, systematic uncertainty and total uncertainty .

HMP 1 HMP 1 RH Temp o

C

% s Ur US UT

o

C

%

1.3328 0.8945 0.0055 0.0037 39.3434 20.8521 2.0000 0.1000 2.0000

HMP 2 HMP 2 RH Temp

0.1001

HMP 3 HMP 3 RH Temp o

C

%

HMP 4 RH

HMP 4 Temp

Pressure Pressure Exthaust Supply

o

Pa

C

%

1.6007 0.0066 -8.0513 0.1000

2.4531 0.0101 16.6964 1.4000

2.5703 0.0106 2.6907 0.2000

3.7454 0.0154 66.7383 2.0000

1.7966 0.0074 10.0383 0.1000

0.4508 0.0019 5.7938 0.1900

1.2758 0.0053 3.9516 0.1900

2.0001

0.1002

1.4000

0.2003

2.0001

0.1003

0.1900

0.1901

kgW/kgM

kgW/kgM kgW/kgM

Exhaust flow Supply flow 3 m3/h kgW/kgM m /h

s Ur US UT

Pa

3.7454 0.0154 33.6857 2.0000

0.3733

0.0120

0.0060

0.0007

0.0008

0.0051

0.0043

0.0146

0.0003

0.0000

0.0001

0.0002 ??

A-17

1.0528

1.6629 0.1096

Membrane Based Heat Exchanger

.

Figure A.11.

Exhaust (red) and supply (blue) air flows through the heat exchanger (based on inlet and outlet values) in a Humiditytemperature diagram. The green lines are relative humidity lines starting at 10% to the left and 100% (saturation line) at the right.

Figure A.12.

Relative humidity (left) and temperature (right) measurements through the test period in all four measurement points.

Figure A.13.

Pressure drop for both air flows. Linear trend line for exhaust air stream is displayed as well. A-18

Appendix.

Figure A.14.

Left: surrounding temperature (red line) and Relative humidity (green line) over the test period. Right: Temperature efficiency.

A-19

Membrane Based Heat Exchanger

.

A.4.3 Experiment 3: DuPont X-membrane A five layer membrane based heat exchanger utilising a membrane from DuPont in this thesis called membrane X was tested. Neither ice formation nor water condensation occurred through the test period.

Table A.3. Data from experiment 3. Standard deviation, random uncertainty, mean value, systematic uncertainty and total uncertainty is shown.

A-20

Appendix.

Figure A.15.

Exhaust (red) and supply (blue) air flows through the heat exchanger (based on inlet and outlet values) in a Humidity-temperature diagram. The green lines are relative humidity lines starting at 10% to the left and 100% (saturation line) at the right.

Figure A.16.

Relative humidity (left) and temperature (right) measurements through the test period in all four measurement points.

Figure A.17.

Pressure drop for both air flows. Linear trend line for exhaust air stream is displayed as well.

A-21

Membrane Based Heat Exchanger

Figure A.18.

.

Left: surrounding temperature (red line) and Relative humidity (green line) over the test period. Right: Temperature efficiency (blue line and moisture transfer efficiency (red line).

A-22

Appendix.

A.4.4 Experiment 4: DuPont PP The poly propylene protection sheet for the DuPont X membrane was tested as a second plastic material. The PP-sheet was used since the material was very similar to the membrane material regarding thickness, colour and behaviour as elasticity etc. The heat exchanger had five layers of plastic sheets. In this experiment the TSI measurement instrument were used to measure the supply air outlet velocity at the duct exit rather than the surrounding temperature and humidity. Condensed water was observed after a short while in the upper exhaust air channel through the acrylic top plate. At the end of the experiment a lot of ice had formed in the exhaust air side. The supply air channels was observed to be narrower caused by the ice layer in the exhaust air side.

Table A.4.

Data from experiment 4. Standard deviation, random uncertainty, mean value, systematic uncertainty and total uncertainty.

A-23

Membrane Based Heat Exchanger

.

Figure A.19.

Exhaust (red) and supply (blue) air flows through the heat exchanger (based on inlet and outlet values) in a Humidity-temperature diagram. The green lines are relative humidity lines starting at 10% to the left and 100% (saturation line) at the right.

Figure A.20.

Relative humidity (left) and temperature (right) measurements through the test period in all four measurement points.

Figure A.21.

Pressure drop for both air flows. Linear trend line for both air streams are displayed as well.

A-24

Appendix.

Figure A.22.

Velocity in one point at the supply air outlet channel exit. Right: Temperature efficiency.

Figure A.23.

Pictures after experiment. Left: Ice was found in the upper right half (near the supply air inlet). Right: side view of the ice and water filled exhaust air side.

A-25

Membrane Based Heat Exchanger

.

A.4.5 Experiment 5: DuPont X A second experiment with the membrane from DuPont was done. Now the supply air flow were greater than the exhaust air flow, opposite of the case in experiment 3. The supply air temperature was about 0oC. Neither ice, condensate water nor crumpling of the membrane were observed in this experiment.

Table A.5.

Data from experiment 5. Standard deviation, random uncertainty, mean value, systematic uncertainty and total uncertainty.

A-26

Appendix.

Figure A.24.

Exhaust (red) and supply (blue) air flows through the heat exchanger (based on inlet and outlet values) in a Humidity-temperature diagram. The green lines are relative humidity lines starting at 10% to the left and 100% (saturation line) at the right.

Figure A.25.

Relative humidity (left) and temperature (right) measurements through the test period in all four measurement points.

Figure A.26.

Pressure drop for both air flows. Linear trend line for exhaust air stream is displayed as well. A-27

Membrane Based Heat Exchanger

Figure A.27.

.

Temperature efficiency (blue line) and moisture transfer efficiency (red line).

A-28

Appendix.

A.4.6 Experiment 6: DuPont X A third experiment with the membrane from DuPont was done. The test conditions were almost the same as for the previous experiment, but the supply inlet temperature was colder. Neither ice, condensate water nor crumpling of the membrane were observed in this experiment.

Table A.6.

Data from experiment 6. Standard deviation, random uncertainty, mean value, systematic uncertainty and total uncertainty.

A-29

Membrane Based Heat Exchanger

.

Figure A.28.

Exhaust (red) and supply (blue) air flows through the heat exchanger (based on inlet and outlet values) in a Humidity-temperature diagram. The green lines are relative humidity lines starting at 10% to the left and 100% (saturation line) at the right.

Figure A.29.

Relative humidity (left) and temperature (right) measurements through the test period in all four measurement points.

Figure A.30.

Pressure drop for both air flows. Linear trend line for exhaust air stream is displayed as well. A-30

Appendix.

Figure A.31.

Temperature efficiency (blue line) and moisture transfer efficiency (red line).

A-31

Membrane Based Heat Exchanger

.

A.4.7 Experiment 7: DuPont X A fourth experiment with the membrane from DuPont was done. Now the test rig was moved closer to the cooling coil and the supply air flow rate was increased to decrease the supply air temperature. The flow rates were not measured, but the supply air flow rate was about three times greater than the exhaust air flow rate. Unfortunately the exhaust air fan stopped working after about 350 min and the result are therefore only displayed for the first 350 min. There were no sign of ice formation in the heat exchanger. A few droplets of condensed water were though found attached to the upper plastic plate. This did however not influence the pressure drop over the heat exchanger.

Table A.7.

Data from experiment 7. Standard deviation, random uncertainty, mean value, systematic uncertainty and total uncertainty is shown.

HMP 1 HMP 1 RH Temp % s Ur US UT

HMP 2 HMP 2 RH Temp

HMP 3 HMP 3 RH Temp

HMP 4 RH

o o o C C C % % % 0.4198 0.1208 0.2205 0.0480 0.8949 0.6092 0.0029 0.0008 0.0015 0.0003 0.0062 0.0042 37.2678 23.2101 41.0420 -10.5020 35.2069 -3.5668 2.0000 0.1000 2.0000 0.1000 1.4000 0.2000

2.0000

0.1000

2.0000 kgW/kgM

0.1000

1.4000

kgW/kgM kgW/kgM

0.2000

HMP 4 Temp

Pressure Pressure Exthaust Supply

o

Pa

C 0.3560 0.0025 2.9113 0.1000

0.2108 0.0015 25.6146 0.1900

0.3547 0.0025 27.2056 0.1900

2.0001

0.1000

0.1900

0.1900

Exhaust flow Supply flow 3 m3/h kgW/kgM m /h

s Ur US UT

0.5993

0.9121

0.0065

0.0007

0.0010

0.0012 ??

??

0.0037

0.0100

0.0004

0.0000

0.0000

0.0001 ??

??

A-32

Pa

2.4474 0.0170 25.3446 2.0000

Appendix.

Figure A.32.

Exhaust (red) and supply (blue) air flows through the heat exchanger (based on inlet and outlet values) in a Humiditytemperature diagram. The green lines are relative humidity lines starting at 10% to the left and 100% (saturation line) at the right.

Figure A.33.

Relative humidity (left) and temperature (right) measurements through the test period in all four measurement points.

Figure A.34.

Pressure drop for both air flows. Linear trend lines for the air streams are displayed. A-33

Membrane Based Heat Exchanger

.

Figure A.35.

Temperature efficiency (blue line) and moisture transfer efficiency (red line).

Figure A.36.

Velocity in one point at the supply air outlet channel exit.

A-34

Appendix.

A.4.8 Experiment 8: DuPont X The membrane based heat exchanger was tested yet another time. The air flow rate on the supply air side was turned down compared to experiment 7. The flow rates were not measured, but the velocity at the supply air outlet was about the half of experiment 7 and at the same level as in experiment 4. The exhaust air humidity were much higher than in experiment 7 (46 to 37% RH). The exhaust air pressure drop experienced a tiny increase compared to the supply air pressure drop. At the end of the test period ice were observed in the exhaust air channels close to the supply air exit. The membrane was also crumbled in this area. No ice was found in the middle exhaust air channel.

Table A.8. Data from experiment 8. Standard deviation, random uncertainty, mean value, systematic uncertainty and total uncertainty is shown.

A-35

Membrane Based Heat Exchanger

.

Figure A.37.

Exhaust (red) and supply (blue) air flows through the heat exchanger (based on inlet and outlet values) in a Humidity-temperature diagram. The green lines are relative humidity lines starting at 10% to the left and 100% (saturation line) at the right.

Figure A.38.

Relative humidity (left) and temperature (right) measurements through the test period in all four measurement points.

Figure A.39.

Pressure drop for both air flows. Linear trend lines for the air streams are displayed. A-36

Appendix.

Figure A.40.

Velocity in one point at the supply air outlet channel exit. Right: Temperature efficiency.

Figure A.41.

Pictures after experiment. Left: The membrane had expanded near the supply air outlet (down to the right in the picture) and can be seen as crumpled. In the same area condensed water had formed and stuck the membrane to the upper acrylic plate of the heat exchanger. Right: the heat exchanger was taken out from the set up. The membrane stretched out again after a few minutes. The oval shaped structures in the lower right side in the picture shows ice that stuck the membrane to the upper plate.

A-37

Membrane Based Heat Exchanger

A.5.

.

Heat Exchanger Prototype: Mechanical Drawing

A-38

Appendix.

A.6.

HSE Report

A-39

Membrane Based Heat Exchanger

.

A-40

Appendix.

A-41

Membrane Based Heat Exchanger

.

A-42

Appendix.

A-43

Membrane Based Heat Exchanger

.

A-44

Appendix.

A-45

Membrane Based Heat Exchanger

.

A-46

Appendix.

A-47

Membrane Based Heat Exchanger

.

A-48

Appendix.

A-49

Membrane Based Heat Exchanger

.

A-50

Appendix.

A-51

Membrane Based Heat Exchanger

.

A-52

Appendix.

A-53

Membrane Based Heat Exchanger

.

A-54